GigaOm Radar for Streaming Data Platforms v6

Event stream processing has become essential to modern data‑driven operations, enabling organizations to act at the pace of real‑time business. Streaming platforms continuously compute over live data as it arrives, providing immediate context about the state of applications, customers, and systems. This capability underpins a wide range of contemporary workloads, from multiagent AI deployments and retrieval augmented generation (RAG) to IoT telemetry, security analytics, financial trading, and clinical monitoring. These use cases depend on the ability to process and interpret data the moment it is generated.

Streaming platforms also play a pivotal role in broader data ecosystems. Their outputs can be enriched with historical data, incorporated into multistage pipelines, and landed in data lakes or lakehouses for downstream analytics. Although many platforms support or complement batch processing, their value lies in surpassing traditional ETL and ELT paradigms by enabling organizations to transform, understand, and act on data instantaneously.

The business impact is significant. Real‑time insight improves decision‑making, strengthens operational efficiency, and enhances customer‑facing experiences. It also accelerates modernization efforts by allowing organizations to extend their existing data investments into real‑time domains without abandoning established architectures. As vendors evolve, they are increasingly evaluated not only on classic streaming capabilities (such as real‑time applicability, windowing, and ease of use) but also on their integration of ML, predictive analytics, and generative AI (GenAI)‑adjacent functionality.

This report evaluates both open source‑based and proprietary streaming platforms, including specialists focused solely on event stream processing and broader data management vendors that offer streaming services or modules. The goal is to help organizations identify solutions that align with their operational requirements, architectural preferences, and long‑term data strategies.

This is our sixth year evaluating the streaming data space. This report builds on our previous analysis and considers how the market has evolved over the last year.

This GigaOm Radar report examines 18 of the top streaming data solutions and compares offerings against the capabilities (table stakes, key features, and emerging features) and nonfunctional requirements (business criteria). It provides an overview of the market, identifies leading streaming data offerings, and helps decision-makers evaluate these solutions so they can make a more informed investment decision.

To help prospective customers find the best fit for their use case and business requirements, we assess how well streaming data solutions are designed to serve specific target markets and deployment models (Table 1).

For this report, we recognize the following market segments:

• Small-to-medium business (SMB): In this category, we assess solutions on their ability to meet the needs of organizations ranging from small businesses to midsize companies. Also assessed are departmental use cases in large enterprises where ease of use and deployment are more important than extensive management functionality, data mobility, and feature sets.

• Large enterprise: Here, offerings are assessed on their ability to support large and business-critical projects. Optimal solutions in this category have a strong focus on flexibility, performance, data services, and features to improve security and data protection. Scalability is another big differentiator, as is the ability to deploy the same service in different environments.

• Specialized: Optimal solutions are designed for specific workloads and use cases, such as big data analytics and high-performance computing (HPC).

In addition, we recognize the following deployment models:

• SaaS: This deployment model is a fully managed option hosted by the vendor. Subscribers to the service pay only for what they use. Vendors are tasked with the significant overhead costs of maintaining resources necessary for organizations to use it and handling other managerial concerns such as updates. An advantage of this deployment model is that large IT teams are not needed to implement it.

• Self-managed: The self-managed option places the burden of managing deployment on the organizations accessing it through the cloud. Those organizations are responsible for provisioning resources for it and managing compute, storage, memory, and other concerns. One of the benefits of this method is that organizations have much more control over the service than they do with the SaaS paradigm, which can help with security and regulatory compliance concerns. Typically, IT personnel are required to make this model effective.

• Public cloud image: With this deployment model, a third-party cloud provider (such as AWS, GCP, or Microsoft Azure) hosts the service. Organizations typically access the cloud solution via containers or, in some cases, virtual machines and can leverage the cloud provider’s underlying compute, memory, and storage, along with scalability resources.

Table 1. Vendor Positioning: Target Market and Deployment Model

Table 1 components are evaluated in a binary yes/no manner and do not factor into a vendor’s designation as a Leader, Challenger, or Entrant on the Radar chart (Figure 1).

“Target market” reflects which use cases each solution is recommended for, not simply whether that group can use it. For example, if an SMB could use a solution but doing so would be cost-prohibitive, that solution would be rated “no” for SMBs.

All solutions included in this Radar report meet the following table stakes—capabilities widely adopted and well implemented in the sector:

• Sophisticated streaming data processing

• Data pipeline simplification

• Data ingestion and processing engine versatility

• Data security

• Basic streaming data quality

Tables 2, 3, and 4 summarize how each vendor in this research performs in the areas we consider differentiating and critical in this sector. The objective is to give the reader a snapshot of the technical capabilities of available solutions, define the perimeter of the relevant market space, and gauge the potential impact on the business.

• Key features differentiate solutions, highlighting the primary criteria to be considered when evaluating a streaming data solution

• Emerging features show how well each vendor implements capabilities that are not yet mainstream but are expected to become more widespread and compelling within the next 12 to 18 months.

• Business criteria provide insight into the nonfunctional requirements that factor into a purchase decision and determine a solution’s impact on an organization

These decision criteria are summarized below.

Key Features

• SQL friendliness: This metric assesses how well an event streaming platform supports SQL for ingesting and processing data. Because SQL is the standard language for most IT teams and developers, this feature enables them to manipulate core aspects of streaming solutions using a familiar, accessible framework.

• Advanced data quality: This feature ensures the quality of data being ingested and processed by streaming platforms. It is a critical means of filtering data streams, shaping them into desired outputs, and making streaming data more manageable for consuming applications.

• Streaming data visualizations: Streaming data visualizations provide an intuitive, graphical way for users to view, interact with, and access data streams. They are an integral component of modern event stream processing solutions because they simplify the user experience.

• Edge applicability: This criterion evaluates how well streaming data solutions support edge deployments. This capability is growing in importance as IoT adoption increases and organizations increasingly process streaming data at the edge rather than sending all of it to centralized cloud environments.

• Advanced analytics: This feature allows organizations to apply sophisticated analytics to incoming data streams. It is essential for real‑time AI applications such as computer vision, GenAI applications, intelligent agents, and geolocation workloads.

• Complex event stream processing: This metric indicates how proficient solutions are at identifying complex events in streaming data for both conventional time‑sensitive applications and emerging ones such as data science or GenAI. These capabilities reflect the core value proposition of streaming platforms: producing meaningful results from complex streaming inputs.

Table 2. Key Features Comparison

Emerging Features

• Vibe coding data pipelines: This feature allows organizations to construct streaming data pipelines using natural language prompts through vibe coding, in which coding agents implement the underlying logic. This capability simplifies pipeline creation even more than visual approaches, making it accessible to a broader range of users.

• Event-driven autoscaling: This feature refines the granularity of autoscaling down to the level of a single event. As such, it is more efficient and cost-effective than alternative forms of autoscaling.

Table 3. Emerging Features Comparison

Business Criteria

• Resilience: This criterion evaluates how effectively event streaming platforms maintain uptime and recover from disruptions. Because streaming data is acted upon immediately, even brief downtime can distort downstream applications and analytics.

• Ease of use: This criterion assesses how easily users can work with the core features of streaming data platforms. Solutions that are simple to operate broaden adoption and democratize usage across both technical and nontechnical roles.

• Flexible scalability: Given the high data volumes inherent in streaming workloads, scalability is essential. This criterion evaluates how vendors deliver scalability and how effectively they support a wide range of deployment patterns.

• Cost effectiveness: This criterion considers how well a platform balances capability and cost, as well as how efficiently it enables organizations to achieve value from their streaming investments.

• Regulatory compliance: This criterion assesses how well solutions help organizations comply with the regulatory requirements that apply to streaming data deployments. Strong compliance capabilities reduce risk and help organizations avoid penalties, reputational damage, and legal exposure.

Table 4. Business Criteria Comparison

The GigaOm Radar plots vendor solutions across a series of concentric rings, with those positioned closer to the center being judged as having the most complete solution. The chart characterizes each vendor on two axes—balancing Maturity versus Innovation and Feature Play versus Platform Play—while providing an arrowhead that projects each solution’s expected evolution over the coming 12 to 18 months.

Figure 1. GigaOm Radar for Streaming Data

As you can see in Figure 1, the vendors are almost evenly distributed on each side of the Radar graphic. This is a testament to the variety of approaches used in this space. There are nearly an equal number of vendors on the Maturity and Innovation halves of the Radar, as well as on the Feature and Platform Play halves. It also alludes to how pervasive the technology underlying streaming data platforms is, since it underscores the reality that vendors are adopting so many different methods to implement it.

However, it’s fairly significant that the majority of the Leaders are on the Innovation half of the Radar. This denotes the efficacy of the copious use of advanced ML, frameworks to empower intelligent agents with the latest streaming data, and tailor-made integrations with AI retrieval systems or vector databases. Still, all the Outperformers are on the Maturity side of the Radar, indicating the fine line between the most progressive vendors and the most steady, dependable ones, especially since the latter are making rapid strides to close that ever-shrinking gap.

In reviewing solutions, it’s important to keep in mind that there are no universal “best” or “worst” offerings; every solution has aspects that might make it a better or worse fit for specific customer requirements. Prospective customers should consider their current and future needs when comparing solutions and vendor roadmaps.

INSIDE THE GIGAOM RADAR

To create the GigaOm Radar graphic, key features, emerging features, and business criteria are scored and weighted. Key features and business criteria receive the highest weighting and have the most impact on vendor positioning on the Radar graphic. Emerging features receive a lower weighting and have a lower impact on vendor positioning on the Radar graphic. The resulting chart is a forward-looking perspective on all the vendors in this report, based on their products’ technical capabilities and roadmaps.

Note that the Radar is technology-focused, and business considerations such as vendor market share, customer share, spend, recency or longevity in the market, and so on are not considered in our evaluations. As such, these factors do not impact scoring and positioning on the Radar graphic.

For more information, please visit our Methodology.

AWS: Amazon Managed Streaming for Apache Kafka and Amazon Kinesis

Solution Overview

Amazon Managed Streaming for Apache Kafka (Amazon MSK) is AWS’s fully managed Kafka service for streaming data ingestion, collection, and processing. It provides native Kafka APIs for both data and administrative operations, along with interfaces for alerting on anomalous events and managing and monitoring Kafka clusters. Amazon Kinesis is AWS’s broader family of streaming data services, used directly by customers and also embedded as an ingestion or delivery layer in third‑party streaming platforms.

Kinesis comprises several services, including Kinesis Video Streams, Kinesis Data Firehose, Kinesis Data Streams, and Amazon Managed Service for Apache Flink. Kinesis Data Streams is a serverless service for building real‑time applications that publish and consume streaming data. Kinesis Data Firehose is a fully managed, serverless delivery service for ingesting, transforming, and loading streaming data into destinations such as Amazon S3 (including Iceberg tables), Amazon Redshift, and Snowflake. Amazon Managed Service for Apache Flink provides access to Flink’s processing and analytics capabilities for stateful stream processing, windowing, and SQL‑based analysis.

AWS is positioned as a Challenger and Fast Mover in the Innovation/Feature Play quadrant of the streaming data Radar chart.

Strengths

AWS scored well on a number of decision criteria, including:

• Complex event processing: Amazon Managed Service for Apache Flink gives customers access to Flink’s constructs for windowing, querying, and joining streaming data. Amazon MSK customers can also use Flink blueprints to vectorize content via Amazon Bedrock without custom code, leveraging Bedrock embedding models, Amazon OpenSearch’s vector store and indexing, and MSK’s streaming data. LangChain integrations streamline chunking workflows. Firehose can deliver data into Iceberg tables stored in Amazon S3, which can then be accessed by Amazon SageMaker Studio for model training and SQL‑based analytics.

• Advanced data quality: Amazon MSK and Kinesis Data Streams integrate with Amazon DataZone to extend schema governance and lineage tracking to streaming data. Organizations can catalog MSK topics, evolve schemas, and monitor schema changes to maintain data quality across streaming pipelines.

• SQL friendliness: Amazon Managed Service for Apache Flink supports ANSI‑compliant Flink SQL for processing streaming data. Users can create tables, run SQL queries, and compute column‑level statistics such as null counts, min/max values, averages, and distinct counts. AWS also provides connectors for Amazon Simple Queue Service (Amazon SQS) and Amazon DynamoDB, broadening the utility of Flink SQL within the platform.

Opportunities

AWS has room for improvement in a couple of decision criteria, including:

• Edge applicability: AWS could strengthen its edge deployment capabilities by expanding native support for processing and filtering streaming data on gateway and edge devices.

• Streaming data visualizations: AWS’s streaming services could improve in this category by enhancing visual tools for comparing historical and streaming data.

Purchase Considerations

Kinesis Data Streams uses a pay‑as‑you‑go pricing model with two options: On‑Demand and Provisioned. In both models, customers pay per gigabyte of data read from and written to streams. The On‑Demand option automatically scales with traffic, while the Provisioned option allows customers to specify capacity in advance.

Use Cases

AWS’s streaming data services are well suited for real‑time data replication for high availability, recommendation engines for media and content platforms, and large‑scale real‑time intelligence for cybersecurity.

Cloudera: Data in Motion

Cloudera’s Data in Motion technology set consists of three chief components: Cloudera Data Flow, Data Streaming, and Edge Management. Cloudera Data Flow relies on Apache NiFi to manage and orchestrate data's movement and universal distribution. Cloudera Streaming is built on a foundation of Apache Flink and Apache Kafka for managing and processing streaming data. Edge Management is based on Apache MiNiFi, an edge agent facilitating NiFi’s features, which delivers management, controls, and monitoring of data within streamlined, resource-constrained edge deployments.

Each of these three Data in Motion building blocks is available on its own as a Kubernetes Operator. Collectively, the full set enables organizations to ingest, process, and analyze low-latency data for immediate analysis and insight. Flink is the chief streaming data processing engine, delivering operational and analytics support. Kafka supplies storage capabilities and is the pub/sub buffer, while NiFi implements lightweight processing at the file level and features robust data distribution capabilities, enhanced by its data pipeline capacity, with more than 450 data source processors.

Cloudera DataFlow provides visual tooling for filtering, cleansing, joining, and enriching data streams, along with dashboards for monitoring job status, logs, and system health. These capabilities support operational visibility across streaming pipelines and Data in Motion workloads.

Cloudera is positioned as a Leader and Fast Mover in the Innovation/Feature Play quadrant of the streaming data Radar chart.

Strengths

Cloudera scored well on a number of decision criteria, including:

• SQL friendliness: Much of Data in Motion’s SQL usability derives from SQL Stream Builder (SSB), Cloudera Streaming’s UI for interfacing with Flink and its APIs via SQL. SSB makes the full complement of Flink’s windowing capabilities (including global, tumbling, sliding, and session windows) accessible through SQL. It also lets users transform data and present it as virtual tables optimized for relational querying. SSB can materialize query results and can be used for both batch and streaming data ingestion.

• Edge applicability: Cloudera Edge Management leverages Apache MiNiFi for streaming data deployments in edge environments. It supports data collection and data management at the edge and features a codeless, drag‑and‑drop UI for building streaming topologies and deploying MiNiFi agents. Using Edge Data Collection, users can deploy lightweight MiNiFi C++ or Java agents that support all of NiFi’s functions. Edge Flow Manager provides a low‑code hub for building, implementing, and tracking deployments of MiNiFi agents at scale.

• Streaming data visualizations: Cloudera DataFlow provides a visual interface for accessing processors to filter, cleanse, join, and enrich data streams. It also offers visualizations for tracking the health of streaming data assets. SQL Stream Builder supplies dashboards to monitor properties, log information, and the status of Data in Motion jobs.

Opportunities

Cloudera has room for improvement in a couple of decision criteria, including:

• Advanced analytics: Improving Cloudera’s built-in capacity to create vector embeddings of streaming data at ingestion time would strengthen the Data in Motion offering.

• Vibe coding data pipelines: Cloudera provides access to external AI models that assist with natural language pipeline creation, where users describe a pipeline and the system generates the underlying code. Expanding native, first-party capabilities in this area would enhance the platform’s usability and increase its value for teams seeking low‑code or AI‑assisted development.

Purchase Considerations

Cloudera’s pricing model is based on compute and data consumption, with compute measured in Cloudera Compute Units, which reflect the number and size of compute nodes in the deployment. For Cloudera Private Cloud deployments, an annual subscription model is available.

Use Cases

Cloudera’s streaming data services support multicloud data movement, edge and IoT deployments, real‑time data products, and pipelines for change data capture (CDC).

Cogility: Cogynt

Cogynt is a unified, real‑time platform that combines data streaming, predictive analytics, statistical AI, and expert system capabilities to deliver low‑latency insights for situational intelligence and decision‑making. It includes a no‑code authoring environment for building real‑time analytics across structured and unstructured data. Its streaming capabilities are supported by integrations with Apache Kafka and Apache Flink, while an integration with Apache Pinot provides a distributed, column‑oriented data store. The platform also supports CDC.

Cogynt uses nonstatistical expert system AI techniques built on hierarchical complex event processing for data transformation. Analytics functions, referred to as “models,” can process data streams continuously without requiring users to issue queries. When querying is needed, users can rely on Apache Pinot or Cogynt’s integration with Apache Superset, an open source framework for data exploration and visualization.

Cogility is positioned as a Challenger and Fast Mover in the Maturity/Platform Play quadrant of the streaming data Radar chart.

Strengths

Cogility scored well on a number of decision criteria, including:

• Streaming data visualizations: Cogynt’s integration with Apache Superset provides a wide range of dashboards and visualizations for exploring data and assessing performance. The platform also includes internal dashboards for Kubernetes, Pinot, Confluent, Redis, PostgreSQL, Istio, and Flink, offering real-time metrics for reliability and performance. Grafana dashboards supply additional visibility into metrics and logs for observability and application performance. Users can author analytics models visually, without code, and the Analyst Workstation enables investigation of data events, including risk history, causal analysis, link analysis, and geospatial views, also without programming.

• Complex event stream processing: Cogynt supports multiple windowing strategies, including late‑arriving, sliding, and tumbling windows. Basic ML capabilities are available, enabling users to deploy risk analysis models, such as Bayesian approaches, to score events across categories such as threats, opportunities, and urgency. Unstructured text processing is enhanced by the Lexicon feature, which filters data using categorization phrases and keywords. The platform’s expert system rules and constraint-based approach provide full explainability for AI-driven findings.

• Advanced analytics: Cogynt includes a range of ML capabilities and also provides nonstatistical rules‑ and constraints‑based AI techniques suitable for streaming data. This combination of statistical and expert system methods is relatively uncommon among streaming data platforms and gives users flexibility in how they model and interpret events.

Opportunities

Cogility has room for improvement in a couple of decision criteria, including:

• Edge applicability: Although Cogynt’s roadmap includes a lightweight version of the platform optimized for edge use cases, this capability is not generally available at the time of publication. Rolling it out would improve the platform’s showing in this criterion.

• SQL friendliness: Cogynt is designed to eliminate the need for programming or SQL, but some organizations may still want the option to interact with additional platform components using SQL for comfort and operational continuity.

Purchase Considerations

Cogynt requires an annual subscription for production, staging, and development deployments within a customer’s VPC. Pricing tiers are based on the number of Entities of Record, which the system continuously monitors, and Entities of Interest (EoI), which are entities related to those being monitored.

Use Cases

Primary use cases include cyberthreat intelligence, social media bot detection, insider threat assessment and management, and pilot air safety intelligence. A significant portion of Cogility’s user base is in the public sector.

Confluent: Confluent Cloud and Confluent Platform

At the end of 2025, Confluent entered into a definitive agreement to be acquired by IBM. The acquisition is intended to strengthen IBM’s real‑time data processing capabilities in support of its intelligent agent and statistical AI strategy.

Confluent’s architecture centers on combining Apache Kafka for data ingestion, storage, and messaging with Apache Flink for querying and stream processing. The platform includes Stream Governance, which provides governance capabilities and a schema registry, along with more than 100 connectors that integrate external sources and sinks with Kafka. Confluent Cloud is a multitenant, fully managed serverless offering powered by Kora, Confluent’s cloud‑native engine built to run Kafka at scale. In recent months, Confluent has released several resources to support AI agent deployments, including Streaming Agents.

Confluent is positioned as a Leader and Fast Mover in the Innovation/Platform Play quadrant of the streaming data Radar chart.

Strengths

Confluent scored well on a number of decision criteria, including:

• SQL friendliness: Confluent customers can use ksqlDB, a Kafka‑native streaming SQL engine, to process streaming data. ksqlDB is accessible through Confluent Cloud, and FlowView provides visualizations of topologies and metrics for ksqlDB workloads. AI Model Inference enables real‑time predictions on streaming data using advanced ML models. Users can also rely on Flink SQL for processing data in Confluent, and built-in ML functions in Flink SQL support applications such as anomaly detection and forecasting.

• Complex event stream processing: Apache Flink is widely recognized for its analytics and stream processing capabilities. Confluent extends these capabilities with Confluent Cloud for Apache Flink, a fully managed, elastically scalable service for joining, enriching, and filtering streaming data. Confluent provides a graphical UI for managing and integrating streaming data, as well as connectors to vector databases such as Zilliz, MongoDB, and Pinecone. With Flink Native Inference, users can run open source AI models directly within Confluent Cloud.

• Streaming data visualizations: Confluent offers a GUI in addition to a REST API and CLI for managing streaming data. Confluent Health+ provides dashboards for monitoring operational metrics and integrates with Prometheus and Grafana. It also supports alerting through Microsoft Teams, email, and Slack. Organizations can use these alerting capabilities to perform anomaly detection based on user‑defined rules.

Opportunities

Confluent has room for improvement in a couple of decision criteria, including:

• Advanced analytics: Confluent provides multiple mechanisms for deploying ML models, including AI Model Inference and Flink Native Inference. Expanding native in‑platform ML capabilities, without relying on external resources, would strengthen these features and broaden the platform’s advanced analytics value.

• Advanced data quality: Stream Governance includes several mechanisms for implementing data quality controls, and the Schema Registry can detect schema drift. More sophisticated features, such as deeper context-aware string analysis, would further enhance the platform’s data quality capabilities.

Purchase Considerations

Organizations can begin using Confluent through a free trial that provides an initial credit allocation. Confluent Cloud uses a consumption-based pricing model in which costs are determined by data ingress and egress, storage, and compute resources for Kafka and Flink. Pricing varies by cluster type (Basic, Standard, or Dedicated) and by the use of additional capabilities such as Stream Governance and fully managed connectors.

Use Cases

Confluent is well suited for real‑time applications such as fraud detection, recommendation systems, and other low‑latency analytics workloads. It also supports digital twin and IoT deployments.

DeltaStream*

DeltaStream is a real‑time, cloud‑native streaming data platform built on Apache Spark, ClickHouse, and Apache Flink, which provides autoscaling and data orchestration capabilities for streaming, real-time, and batch analytics. DeltaStream’s 2025 addition of Spark gives the platform lakehouse functionality. The platform supplies a unified layer for data ingestion, data delivery, data serving, transformation, stateful processing, data governance, and storage of streaming data. A central component of the offering is a streaming SQL engine that executes continuous queries and applies streaming semantics using ANSI SQL with DeltaStream extensions. There’s a single SQL interface for workloads involving ClickHouse, Flink, and Spark.

Via its ClickHouse underpinnings, DeltaStream generates materialized views of streaming data that support analytics, enrichment, and operational use cases. Organizations can build streaming data pipelines using a broad set of connectors for sources such as Apache Kafka, Amazon Kinesis, and CDC systems, as well as sinks including ClickHouse, data warehouses, and object stores. A unified backplane provides data security, metadata management, compute orchestration, and data cataloging, all accessible through a common UI and API layer.

DeltaStream is positioned as a Challenger and Outperformer in the Maturity/Platform Play quadrant of the streaming data Radar chart.

Strengths

DeltaStream scored well on a number of decision criteria, including:

• SQL friendliness: DeltaStream’s engine relies on Apache Flink SQL (based on ANSI SQL) with DeltaStream extensions optimized for streaming semantics and use cases. As a result, users can employ SQL to work with both data at rest and data in motion. SQL can be used for deduplication, aggregation, enrichment, and advanced stateful processing such as windowing and joins. Organizations can also use SQL to define event time semantics and generate continuous materialized views.

• Complex event stream processing: DeltaStream supports a wide range of time‑windowing options, including session, sliding, and tumbling windows, which can be combined as needed. The platform also supports stream‑to‑stream joins and subqueries, all accessible through SQL. User‑defined functions (UDFs) enable users to generate vector embeddings through integrations with vector stores such as PostgreSQL’s pgvector extension. UDFs are also useful for implementing low‑latency inference pipelines that rely on third‑party ML services, including those with generative models.

• Advanced data quality: DeltaStream uses Flink to support several aspects of data quality. It enables RegEx‑based processing, rule‑based transformations for standardizing values such as dates, and schema enforcement. The platform also includes capabilities for handling late‑arriving events and events that arrive out of chronological order.

DeltaStream is classified as an Outperformer because of the substantial advancements it is projected to make as a low-latency context layer for AI agents.

Opportunities

DeltaStream has room for improvement in a couple of decision criteria, including:

• Edge applicability: DeltaStream currently supports a Kubernetes‑based bring-your-own-cloud (BYOC) deployment model. The vendor could improve its score in this area by offering a dedicated product or service for edge deployments.

• Streaming data visualizations: DeltaStream provides dashboards for inspecting streaming data, pipeline health, metrics for materialized views, and watermarking. These visualizations could be strengthened by adding mechanisms to compare historical and streaming data through a single point‑and‑click interface.

Purchase Considerations

DeltaStream is available as an integrated SaaS offering. Organizations can also deploy it through a BYOC model via major cloud marketplaces.

Use Cases

DeltaStream positions itself as a real‑time layer for supplying context to intelligent agents and GenAI systems. It also supports more traditional streaming workloads such as continuous transformations, CDC pipelines, and low‑latency enrichment, giving teams a unified way to build operational and AI‑driven applications on top of event data.

EsperTech: Esper and NEsper

Esper and NEsper are open source complex event processing and streaming data analytics offerings available via the GNU General Public License (GPL). Esper is designed for Java and Java Virtual Machines, while NEsper works with .NET products. Esper Enterprise and Esper HA are closed source offerings; Esper HA delivers high availability, while Esper Enterprise scales horizontally and deploys in clouds and data centers. Organizations can integrate Esper and NEsper as libraries into their Java or .NET applications. Esper Enterprise and Esper HA can deploy as standalone, server-like products that support containers. They can also be incorporated as part of existing Kafka applications.

EsperTech is positioned as a Challenger and Fast Mover in the Maturity/Platform Play quadrant of the streaming data Radar chart.

Strengths

EsperTech scored well on a number of decision criteria, including:

• Edge applicability: Because Esper and NEsper can be embedded as lightweight Java and .NET libraries, they naturally support edge deployments. They can run on devices that support JVM or .NET runtimes, and they offer containerized deployment options through Docker images and Kubernetes orchestration. Their small footprint (typically under 32 MB of RAM) allows them to run on a wide range of CPUs, including diskless environments. EsperTech also provides extensive documentation and APIs for extending and embedding the engines in edge applications.

• SQL friendliness: Esper Enterprise builds on Kafka and benefits from Kafka’s horizontal scalability and masterless, partitioned architecture. The platform dynamically assigns partitions to available resources and leverages Kafka Streams-style processing for distributed execution. Esper and NEsper use Event Processing Language (EPL), a SQL‑like language that supports a range of data quality operations, including filtering, cleansing, schema definition, pattern matching, and regular expressions. EPL’s compiler and runtime further enhance performance by optimizing query planning, CPU utilization, and memory efficiency.

• Advanced streaming data quality: EPL supports a range of data quality operations, including filtering, cleansing, schema definition, pattern matching, and regular expressions. Users can also implement conditional logic and other conventions directly in EPL.

Opportunities

EsperTech has room for improvement in a couple of decision criteria, including:

• Advanced analytics: Although Esper is extensible and integrates with third‑party analytics frameworks, it lacks native constructs for advanced analytics, including support for embedding vector stores or other modern AI/ML components.

• Streaming data visualizations: Esper provides a GUI with point‑and‑click functionality, but users must rely on EPL for filtering, transformation, and other processing tasks. These capabilities are not exposed through the GUI, limiting no‑code and low‑code usability.

Purchase Considerations

Esper and NEsper are available under the GPLv2 open source license, which includes the runtime and core streaming analytics compiler. Commercial support is available for production and development environments. Independent software vendors can redistribute Esper through an OEM agreement with associated licensing fees. Customers seeking the full platform (including horizontal scalability and high availability) can license Esper Enterprise and Esper HA under either an unlimited subscription model or a per‑process subscription model.

Use Cases

Esper is well suited for high-throughput, real-time applications across domains such as finance, cybersecurity, and telecommunications. Common scenarios include situational awareness, event correlation, and analytics involving time-based or multistream event processing.

Google Cloud: Pub/Sub and Dataflow

Dataflow, built on open source Apache Beam, provides both batch and streaming data processing for Google Cloud customers. Pub/Sub is a fully managed, asynchronous messaging service for topic‑based data ingestion. Google Cloud users can also access Google Cloud Managed Service for Apache Kafka. Pub/Sub commonly serves as the messaging backbone for Dataflow pipelines and can ingest streaming data from external services such as Confluent Cloud, Amazon MSK, and Azure Event Hubs. Dataflow also supports Tensor Processing Units (TPUs), Google’s custom accelerators for large‑scale ML and AI workloads.

Google Cloud is positioned as a Challenger and Fast Mover in the Innovation/Feature Play quadrant of the streaming data Radar chart.

Strengths

Google Cloud scored well on a number of decision criteria, including:

• SQL friendliness: Dataflow supports SQL through Beam SQL, enabling users to author and query pipelines using SQL syntax. Beam SQL allows organizations to work with both bounded and unbounded data within the Beam programming model. Pub/Sub integrates with multiple streaming engines, including Apache Flink and Apache Kafka, allowing users to leverage SQL‑based tooling such as Flink SQL and ksqlDB when building downstream processing pipelines.

• Streaming data visualizations: Pub/Sub and Dataflow provide several capabilities for monitoring and visualizing streaming workloads. Pub/Sub integrates with OpenTelemetry for tracing client library operations and supplies audit logs that track resource creation, modification, and deletion. Dataflow offers data lineage, configurable dashboards for pipeline metrics, and data sampling features that enhance observability at any stage of a pipeline. Cost‑monitoring tools are also available.

Opportunities

Google Cloud has room for improvement in a couple of decision criteria, including:

• Advanced data quality: Expanding the built-in data quality features of Dataflow and Pub/Sub (such as semantic validation, pattern detection, or entity classification) would strengthen Google Cloud’s position in this criterion.

• Edge applicability: Although Dataflow and Pub/Sub can interface with edge deployments via Google Distributed Cloud Edge, the need for this additional service makes such deployments less streamlined. Providing more native, integrated edge‑processing features would improve Google Cloud’s score in this area.

Purchase Considerations

Dataflow is billed per second, with hourly rates prorated to reflect actual usage. Pub/Sub pricing is based on message storage, message delivery throughput, and data transfer across regions or zones within Google Cloud.

Use Cases

Google’s streaming services support IoT‑centric scenarios such as manufacturing telemetry, predictive and preventive maintenance, and digital twin applications.

Hazelcast: Hazelcast Platform

The Hazelcast Platform combines a fast in‑memory data store with a distributed compute engine (Hazelcast Engine) in a single runtime. Hazelcast Engine powers distributed caching and stream processing (via the Jet engine) for scalable, cloud‑native performance, including Kubernetes support, enabling real‑time action on low‑latency data. The in‑memory data grid provides streaming data topics and messaging capabilities. Hazelcast Engine is accessible through SQL and more than 180 connectors and APIs. The platform also includes real‑time ML inference via ML Inference Runner, capabilities for a feature store, vector search functionality, and Geo/WAN replication.

Hazelcast is positioned as a Leader and Fast Mover in the Maturity/Platform Play quadrant of the streaming data Radar chart.

Strengths

Hazelcast scored well on a number of decision criteria, including:

• Complex event stream processing: Hazelcast’s distributed cache can function as a feature store, allowing organizations to materialize feature values into Hazelcast clusters and update them with real‑time streaming data. These capabilities support low‑latency model training and fine tuning. Hazelcast provides a range of standard aggregations, such as maximum, minimum, average, and standard deviation, as well as additional advanced aggregation functions. Time‑based windowing options include tumbling, sliding, session, and global windows. Additional temporal controls allow users to specify how long event data should be retained.

• Advanced analytics: ML Inference Runner enables real‑time predictions on streaming data using models written in Python, C++, and Java. It can operate within virtual Python environments, and fault tolerance mechanisms ensure models are redeployed to alternate resources in the event of failures. Hazelcast’s vector search capability, Vector Collection, is currently in beta and allows enterprises to store, manage, and search embeddings and associated metadata. Organizations can build pipelines that incorporate their embedding model of choice.

• SQL friendliness: Hazelcast allows organizations to access, transform, and query both streaming and stored data using SQL. Users can join streaming data with data at rest in a single query. Supported SQL‑queryable structures include streaming topics, JSON documents, and Hazelcast maps, which are distributed key‑value data structures.

Opportunities

Hazelcast has room for improvement in a couple of decision criteria, including:

• Advanced data quality: Enhancing Hazelcast’s native data quality features (such as support for more expressive regular expression handling and additional validation functions) would strengthen its capabilities in this area.

• Vibe coding data pipelines: Providing built‑in models and tooling for natural language‑driven pipeline creation would simplify the development of streaming topologies and improve usability for nonspecialist users.

Purchase Considerations

Hazelcast Platform is available via an annual subscription license for the self‑managed software. The managed service offering, Hazelcast Cloud, can be purchased through prepaid credits or a pay‑as‑you‑go model.

Use Cases

Hazelcast is widely used in financial services for real‑time recommendations, personalized offers, fraud detection, and other latency‑sensitive workloads. More broadly, it is employed for distributed caching and high‑performance in‑memory computing across a range of horizontal use cases.

IBM: StreamSets, Cloud Pak for Data, Event Streams for IBM Cloud

IBM provides a set of services that support streaming data deployments. IBM StreamSets specializes in data ingestion and transformation, integrates with Apache Pulsar and Apache Kafka, enables CDC, and supports both ETL and ELT for batch ingestion. IBM Event Streams is available through IBM Software Hub and integrates with watsonx.data. It ingests event data into IBM Cloud via a Kafka‑based messaging bus. The service integrates with Kafka client APIs, Kafka Connect, and Kafka Streams and provides capabilities for landing data streams into downstream sinks. IBM Cloud Pak for Data is a collection of integrated software modules for implementing a data fabric architecture. Key components include watsonx.ai Studio for building and deploying AI models and IBM Data Virtualization, which provides query‑in‑place and virtualization capabilities.

IBM is positioned as a Challenger and Fast Mover in the Innovation/Feature Play quadrant of the streaming data Radar chart.

Strengths

IBM scored well on a number of decision criteria, including:

• Complex event stream processing: StreamSets provides operators for a variety of streaming data processing functions. Users can specify rolling and sliding windows and access standard aggregations such as count, sum, minimum, maximum, and average, along with support for more advanced aggregation functions. Filtering and grouping capabilities are available, allowing users to organize data by expressions before applying calculations. By using IBM’s frameworks for building and deploying ML models, organizations can orchestrate pipelines that train models on streaming data.

• Advanced analytics: watsonx.ai Studio provides resources for accessing, fine‑tuning, and deploying large language models (LLMs) and can be used within Cloud Pak for Data pipelines that incorporate streaming data. In addition to model libraries, watsonx.ai Studio supplies workflows, runtimes, and APIs for these deployments. Users can build pipelines that generate vector embeddings from streaming data. watsonx.data includes a Milvus‑based vector database, and StreamSets can land data into this store.

• Edge applicability: IBM supports edge deployments for its streaming data services through several mechanisms. StreamSets Data Collector Edge can run pipelines on edge devices via a lightweight execution engine. IBM Edge Computing for Devices, built on the Open Horizon project, enables users to manage edge nodes and reduce deployment risk.

Opportunities

IBM has room for improvement in a couple of decision criteria, including:

• Advanced data quality: IBM’s native data quality capabilities for streaming data are not as strong as those of the leaders in this space. Strengthening these features would help close the gap.

• Streaming data visualizations: Adding features that enable single‑click, visual comparisons between streaming and historical data would improve IBM’s competitiveness in this criterion.

Purchase Considerations

Prospective customers can evaluate IBM’s streaming data services under a 30‑day free trial. Pricing for IBM StreamSets is based on virtual processing cores (VPCs), with customers charged $1,050 per VPC per month.

Use Cases

IBM’s streaming data services support use cases such as monitoring social media streams for engagement analysis, equipment asset monitoring, predictive maintenance, and market tracking scenarios in financial services.

Informatica (Salesforce): Informatica Intelligent Data Management Cloud

Informatica Intelligent Data Management Cloud (IDMC) is a comprehensive data management, integration, and governance platform that provides a range of streaming data capabilities, including data ingestion, data replication, event processing, and data integration. Salesforce acquired Informatica in 2025. IDMC supports processing of real‑time, batch, and event‑driven data and accommodates incremental, initial, and combined load types. Data replication is accessed through a self‑service wizard with multiple CDC options (including audit mode and soft delete) for ingesting data from, or copying it to, messaging hubs, data lakes, and data warehouses. Continuous event processing is available through Data Engineering Streaming, an Informatica service built on Spark Streaming that supports both batch and streaming workloads. IDMC customers can also use a variety of stream‑processing engines for cloud ingestion and event processing.

Informatica is positioned as a Challenger and Fast Mover in the Maturity/Feature Play quadrant of the streaming data Radar chart.

Strengths

Informatica scored well on a number of decision criteria, including:

• Advanced data quality: Users can implement real‑time data quality checks through REST APIs to evaluate, cleanse, and enrich data. Supported features include automatic schema evolution for AI‑powered parsing of semistructured and unstructured data. Data Engineering Streaming includes prebuilt transformations for standardization, address validation, and classification. IDMC also provides dynamic schema capabilities, such as refreshing a data object’s structure when parameterized dynamic mapping jobs complete. Informatica’s AI engine, CLAIRE, can identify and remediate schema drift while determining data structures and suggesting next actions.

• Advanced analytics: Although IDMC does not include its own vector store, it supports common vector‑related tasks such as generating embeddings and chunking content. The platform offers ready‑made workflows (“recipes”) for LLM use cases, including those that incorporate intelligent agents, across model providers. Informatica’s extensive integrations with streaming data platforms (including Apache Kafka, Apache Flink, Confluent Kafka, Amazon Kinesis, and Azure Event Hubs) support multiple windowing techniques. EMR‑based deployments can also be used when configured with streaming data processing engines such as Spark Streaming or Flink. Integrations with services such as Amazon SageMaker Lakehouse and Amazon Bedrock enable organizations to build pipelines that perform model scoring and inference on both streaming and batch data.

• Edge applicability: With Informatica Edge Data Streaming, organizations can collect and aggregate machine data at the edge, process it, and write it to downstream sinks in real time. This distributed, highly available engine allows users to filter and transform data on devices and edge gateways. Supported sources include MQTT brokers, event logs, HTTP sources, and call detail records.

Opportunities

Informatica has room for improvement in a couple of decision criteria, including:

• Streaming data visualizations: Expanding the number and capabilities of options for visualizing streaming data would strengthen Informatica’s position in this criterion.

• Complex event stream processing: Increasing Informatica’s native support for frameworks such as ASTORE and ONNX (both for training prescriptive and predictive models on streaming data and for scoring their results) would enhance its capabilities in this area.

Purchase Considerations

IDMC’s pricing model allows customers to select the services they wish to use. Pricing is based on Informatica Processing Units (IPUs), with consumption determined by factors such as compute usage, number of events processed, and number of connections.

Use Cases

Streaming data use cases supported by Informatica include industrial internet and manufacturing applications, cybersecurity, reservation and customer information tracking in hospitality and travel, and IoT deployments in healthcare.

Materialize

Solution Overview

Materialize characterizes itself as a live operational layer that sits between upstream data sources and downstream applications. It incrementally maintains materialized views defined in SQL, allowing organizations to express business logic once and have the results continuously updated as new data arrives. Materialize provides mechanisms for organizations to define, complete, and quickly pull up canonical business objects impacted by the latest data. The platform ingests CDC streams and other event data from systems such as Kafka and OLTP databases. Its PostgreSQL-compatible interface enables users to define and query derived views using standard SQL. Because Materialize processes only the changes rather than recomputing full results, applications can query always-fresh views without incurring the cost of repeated transformations.

SQL is the primary interface for defining transformations, expressing business logic, and querying derived states. Views can be indexed and queried directly through the SQL interface (a pull model), or Materialize can publish incremental updates to downstream systems such as Kafka topics or Iceberg tables (a push model).

Materialize is positioned as a Challenger and Outperformer in the Maturity/Platform Play quadrant of the streaming data Radar chart.

Strengths

Materialize scored well on a number of decision criteria, including:

• Complex event stream processing: Materialize supports a range of SQL-based stream-processing capabilities, including fixed (tumbling) and sliding windows, time-bucketed aggregations, joins, subqueries, and common table expressions (CTEs). Because Materialize incrementally maintains derived views, many workloads that would traditionally require explicit windowing can instead be modeled as continuously updated tables representing the current state of an entity or process. The platform supports recursive CTEs, nested subqueries, and complex SQL compositions, enabling expressive modeling of streaming and relational logic.

• SQL friendliness: SQL is the core mechanism for transforming, joining, and querying data in Materialize. The query optimizer plans SQL statements as incremental dataflows that react to changes in upstream inputs. PostgreSQL-compatible syntax allows users to join CDC streams from multiple operational systems to construct unified business entities composed of events from different sources. SQL can also be used to implement data quality logic such as validation rules, deduplication, and filtering.

• Advanced data quality: Organizations can use SQL to enforce business rules, remove duplicates, filter events, and implement temporal logic for event-time handling, including time-based windows. These capabilities allow teams to embed data quality controls directly into the incremental views that power downstream applications.

Materialize was classified as an Outperformer due to its capacity for helping organizations rely on SQL while defining canonical business objects and rapidly updating them with the latest available data.

Opportunities

Materialize has room for improvement in a couple of decision criteria, including:

• Edge applicability: Materialize is well suited for maintaining real-time derived views of distributed operational systems. Extending the platform with more traditional edge processing capabilities would strengthen its position in this category.

• Advanced analytics: Several competitors offer built-in ML and AI integrations, including vectorization of streaming data. Adding features in this domain would strengthen Materialize’s position relative to competitors.

Purchase Considerations

Materialize is available as a fully managed SaaS offering (Materialize Cloud) or as self-managed software under commercial or free licenses. Materialize Cloud follows a usage-based pricing model, billed monthly for storage, network, and compute consumption or through annual prepaid commitments. Self-managed deployments are available under a free community license or a paid enterprise license.

Use Cases

Materialize supports use cases such as powering AI and automated systems with continuously updated structured context, delivering real-time insights into operational systems and enabling low-latency, customer-facing applications that rely on continuously updated data.

Microsoft: Azure Event Hubs, Azure Stream Analytics, Azure Data Explorer, and Fabric Real-Time Intelligence

Solution Overview

Microsoft Fabric’s Real‑Time Intelligence (RTI) brings together Microsoft’s streaming and event‑driven capabilities within Fabric through a unified user experience and shared governance SaaS experience. RTI builds on three core Azure services—Azure Event Hubs for event ingestion, Azure Stream Analytics (ASA) for real‑time processing, and Azure Data Explorer (ADX, originally code‑named Kusto) for time series and log analytics—and brings them into Fabric as a cohesive platform. Eventstreams, a no‑code and code‑optional abstraction layer, simplifies the creation of streaming pipelines by providing a unified surface for connecting sources, ingesting events, applying transformations, and routing data to downstream targets. These targets include Fabric Lakehouses and Eventhouses, the latter being a serverless analytics engine built on ADX technology accessible through SQL or Kusto Query Language (KQL). RTI also integrates with Fabric IQ, a unified context layer that models business entities and their relationships across real-time data and data-at-rest. Fabric IQ is an emerging, preview‑stage capability within Microsoft Fabric.

Microsoft is positioned as a Leader and Fast Mover in the Maturity/Feature Play quadrant of the streaming data Radar chart.

Strengths

Microsoft scored well on a number of decision criteria, including:

• Streaming data visualizations: RTI provides low‑code and no‑code mechanisms for ingesting, transforming, analyzing, and acting on streaming data. Users can build Real‑Time Dashboards in Fabric directly with KQL, independent of Power BI, while Azure Stream Analytics can output directly to Power BI for real‑time visualization. KQL can also be used through Notebooks in Fabric for Python‑based visualizations. Azure Stream Analytics includes a codeless editor for point‑and‑click transformations over Event Hubs data, and visualizations from the Azure portal can also be accessed through ASA for operational monitoring.

• Complex event stream processing: Azure Stream Analytics delivers low‑latency analytics for data from Azure IoT Hub, Azure Event Hubs, Azure Blob Storage, Azure Data Lake Storage (ADLS) Gen2, and Apache Kafka. It supports sliding, hopping, session, and tumbling windows, along with aggregations, geospatial functions, snapshots, array operations, and date‑time transformations. These capabilities extend into Fabric through Eventstreams, which provides a simplified, visual, low-code/no-code interface for building and managing streaming pipelines without requiring direct interaction with underlying Azure services. Eventstreams also supports connectors to many more input sources, including Google Cloud Pub/Sub, Amazon Kinesis, MQTT, PostgreSQL CDC, MongoDB CDC, Azure SQL CDC, and CosmosDB CDC. Activator is an RTI component that enables no-code, real-time streaming triggers that can not only send alerts but also invoke actions and workflows.

• SQL friendliness: Eventhouses support both SQL and Kusto Query Language (KQL) for querying real‑time and historical data. RTI enables organizations to join and enrich streaming data using SQL‑based transformations within Eventhouses, while Azure Stream Analytics provides a SQL dialect based on T‑SQL, extended with streaming constructs for windowing and real‑time processing.

Opportunities

Microsoft has room for improvement in a couple of decision criteria, including:

• Advanced data quality: Data quality can be implemented using SQL expressions and Azure Schema Registry for schema enforcement. Eventhouse also enforces table schemas upon ingesting events. Customers can utilize update policies and materialized views, which process events as they are ingested, to perform data validation. RTI enables users to define Event Schema Sets and use them in Eventstreams to validate events and drop events that fail schema validation (and log them as errors). Adopting more measures for pinpointing the context of strings would bolster the vendor in this category.

• Vibe coding data pipelines: RTI offers interactive Data Agents to answer natural language queries and Operations Agents for autonomous and semi-autonomous monitoring and proactive action. Enhancing the ability to generate streaming topologies through natural language prompts (automating more of the underlying code and configuration) would improve usability for teams adopting GenAI-assisted development.

Purchase Considerations

Pricing for Azure Event Hubs depends on capacity, retention period, geographic zones, and the volume of ingress events. When used through Microsoft Fabric, RTI workloads run on Fabric capacities with no separate Azure billing, provisioning, or subscription required.

Use Cases

Microsoft’s streaming capabilities support a wide range of real‑time applications. RTI includes features for building and operating digital twins, and it can update intelligent agents with real‑time signals through Fabric IQ’s emerging semantic and grounding capabilities. RTI also includes Graphs and Maps capabilities for customers to perform advanced graph analytics and visualize geospatial maps over real-time data in RTI. Additional use cases include real‑time recommendations, personalization systems, operational analytics, event‑driven e‑commerce solutions, anomaly detection, graph analytics, threat analysis, fraud detection, and supply chain management.

Oracle: Oracle Cloud Infrastructure GoldenGate Stream Analytics 26ai*

Solution Overview

Oracle GoldenGate Stream Analytics 26ai (GGSA) is a streaming analytics service that complements Oracle GoldenGate by enabling real-time processing, enrichment, and analysis of event data. Delivered as a managed, cloud-native service on Oracle Cloud Infrastructure (OCI), it provides a low-code user experience based on the Oracle Redwood design system. GGSA supports AutoML, integration with ML models, and the ability to work with vector embeddings. It can stream data into Oracle AI Database 26ai by writing to vector columns, and it integrates with services such as HeatWave GenAI and OCI Generative AI Agents.

The service also integrates with GoldenGate 26ai Data Stream, an AsyncAPI-based pub/sub interface for creating new data streams and connecting to existing ones. Data Stream allows users to consume event changes without requiring Kafka clusters or additional GoldenGate instances. OCI GoldenGate continues to support Kafka for data ingestion and for transporting data between sources and targets.

Oracle is positioned as a Leader and Fast Mover in the Innovation/Feature Play quadrant of the streaming data Radar chart.

Strengths

Oracle scored well on a number of decision criteria, including:

• Complex event stream processing: GGSA supports importing ML models via Open Neural Network eXchange (ONNX) and Predictive Model Markup Language (PMML), and it can access models from Oracle Machine Learning and AutoML to score streaming events. AutoML reduces the effort required to select and configure models for common use cases. The service includes native functions for sliding and tumbling windows, geospatial analytics, time series analysis, and related operations. GGSA can stream data to vector stores by writing to vector columns in Oracle AI Database 26ai or by directing streams to third-party vector engines. It also integrates with HeatWave GenAI. Organizations can additionally use OCI Generative AI Agents, which can be integrated with streaming pipelines, to apply agent-based patterns to streaming data, including RAG, agent planning, ranking and reranking of retrieved content, and guardrails and grounding mechanisms designed to reduce hallucinations.

• SQL friendliness: Organizations can implement logic for processing streaming data with Continuous Query Language (CQL), Oracle’s proprietary SQL-like language for continuous queries. With CQL, users can specify filtering, aggregation, and windowing operations, creating temporal views over streaming data that behave similarly to relational tables. CQL expressions can also be customized to encode specific business rules and use case-specific logic.

• Streaming Data Visualizations: Oracle GoldenGate Veridata, a data synchronization, data comparison, and data repair solution, contains mechanisms for monitoring data, some of which involve visual approaches. Additionally, GGSA has constructs for visually implementing facets of analytics; it’s possible to access real-time dashboards for more intuitive understanding of streaming data developments. GGSA also includes no-code, visual functionality for enriching and filtering streaming data and reference data.

Opportunities

Oracle has room for improvement in a couple of decision criteria, including:

• Advanced analytics: Oracle’s streaming data services already provide multiple mechanisms for implementing advanced analytics, including ML model integration and time series functions. Adding more built-in transformations tailored to common industry patterns (such as IoT feature extraction or time series feature engineering) would further strengthen its position in this area.

• Advanced data quality: Oracle customers can access a standalone data quality module and apply ML techniques to score the quality of incoming data. Enhancing the platform’s ability to support more advanced data quality checks, such as semantic validation or context-aware string handling, would make Oracle even more competitive on this criterion.

Purchase Considerations

Pricing for Oracle Cloud Infrastructure GoldenGate Stream Analytics 26ai is based on Oracle Compute Processing Units, which reflect the compute resources consumed by specific deployments.

Use Cases

Oracle’s streaming data services are designed to complement its broader portfolio of AI, LLMs, and intelligent agents. They support a wide range of streaming use cases, including IoT telemetry processing, digital twins, real-time recommendations, and event-driven integrations with Oracle databases and applications.

SAS: Event Stream Processing

Solution Overview

SAS Event Stream Processing (ESP) is a distributed in‑memory platform for building, deploying, and operationalizing continuous queries and streaming data pipelines. The platform is composed of four primary components: ESP Server, which executes streaming projects; ESP Client, which communicates with the server to submit and manage jobs; SAS Event Stream Processing Studio, a visual environment for designing, testing, and deploying streaming pipelines; and SAS Event Stream Processing Analytics, which provides advanced analytics and ML functions for in‑stream scoring and transformation.

Organizations can connect to and consume streaming data at scale, apply transformations, and score events in real time. ESP supports importing analytics models from SAS Model Manager as well as models packaged in ONNX. A scoring API enables on‑demand evaluation, allowing streaming data to be scored in real time using pretrained models.

SAS is positioned as a Challenger and Fast Mover in the Innovation/Platform Play quadrant of the streaming data Radar chart.

Strengths

SAS scored well on a number of decision criteria, including:

• Advanced analytics: SAS ESP integrates with SAS Model Manager, enabling organizations to deploy a wide range of pretrained ML and predictive models for real‑time scoring. Supported analytics include regression, classification, anomaly detection, time series analysis, and selected computer vision and natural language processing functions. ESP provides operators for feature extraction, text processing, and image preparation, enabling real-time scoring of streaming data with minimal preprocessing. Models can be imported from SAS Model Manager or from ONNX, enabling interoperability with frameworks such as PyTorch and TensorFlow.

• Complex event stream processing: ESP offers a broad set of windowing constructs (including sliding, tumbling, and session windows) for aggregations, temporal analysis, and pattern detection. The platform supports in‑stream scoring through an analytics store, a scoring API, and model zoo integrations that allow organizations to deploy pretrained deep learning models. ONNX‑based execution can leverage GPUs and hardware acceleration through NVIDIA CUDA and TensorRT, improving performance for supported workloads.

• Advanced data quality: SAS provides built-in data quality functions that can be applied during ingestion and processing. These include mechanisms for standardizing, cleansing, categorizing, and validating data, as well as detecting anomalies or unexpected changes in streaming inputs. Text processing, statistical checks, and other quality controls can be applied directly within ESP pipelines to ensure downstream analytics receive reliable data.

Opportunities

SAS has room for improvement in a couple of decision criteria, including:

• SQL friendliness: Expanding native SQL support for defining transformations and interacting with ESP pipelines would make the platform more accessible to teams that rely heavily on SQL for data engineering and analytics.

• Event‑driven autoscaling: While SAS ESP scales effectively through conventional deployment methods, adding more robust event‑driven autoscaling capabilities would improve elasticity and operational efficiency for dynamic workloads.

Purchase Considerations

SAS pricing is tailored to specific customer requirements and deployment scenarios. The vendor offers free trials and demonstrations that organizations can evaluate prior to receiving a customized quote.

Use Cases

SAS ESP is well suited for advanced analytics and data science use cases that require real‑time scoring or continuous model refinement. Common applications include streaming feature extraction, real‑time model scoring, anomaly detection, computer vision preprocessing, and natural language processing applied to event streams.

StreamNative: ONE StreamNative Platform

Solution Overview

The ONE StreamNative Platform is a cloud‑native offering for real‑time messaging, event streaming, and AI agent orchestration. It is powered by three engines: Ursa, a leaderless streaming engine introduced in 2025; the classic Apache Pulsar engine; and Orca Agent Engine, a preview‑stage runtime for orchestrating AI agents. Ursa provides Apache Kafka compatibility through Kafka APIs and writes directly to Delta Lake and Apache Iceberg tables without data copying or reliance on broker disks, including support for Availability Zone replication. It integrates with Databricks Unity Catalog, AWS S3 tables, and Snowflake’s Open Catalog.

The Pulsar engine uses Apache BookKeeper for messaging and supports MQTT, Pulsar, and Kafka protocols (via Kafka‑on‑StreamNative, which enables Kafka workload migration without rewriting client code). Orca Agent Engine supplies an event‑driven runtime for building and scaling AI agents, supporting Google’s Agent Development Kit (ADK), LangGraph, and the OpenAI Agents SDK. StreamNative also provides an open source MCP Server for agent connectivity. The platform includes Universal Connectivity (UniConn), a connector framework with more than 200 connectors spanning Pulsar IO and Kafka Connect ecosystems.

StreamNative is positioned as a Leader and Fast Mover in the Innovation/Platform Play quadrant of the streaming data Radar chart.

Strengths

StreamNative scored well on a number of decision criteria, including:

• Complex event stream processing: StreamNative supports fixed and nonoverlapping tumbling windows and sliding windows (including overlapping intervals with configurable slide durations) through Pulsar Functions, its serverless compute engine. More advanced constructs, such as both conventional and activity‑based session windows, are available through Apache Flink. With Flink SQL, organizations can perform interval and windowed inter‑stream joins, temporal table joins for point‑in‑time lookups, and stream‑to‑table joins. Subquery capabilities include embedding SELECT statements in JOIN clauses, FROM clauses, and WHERE conditions. ML models can be trained and scored through Spark MLlib and Flink SQL integrations, while Pulsar Functions can load pretrained models to score events in real time. Pulsar Functions can also generate vector embeddings via APIs for Cohere, OpenAI, and self‑hosted models, and users can access embeddings stored in lakehouse tables. Connectors are available for streaming data to vector stores such as Qdrant, Weaviate, and Pinecone.

• Edge applicability: StreamNative Private Cloud supports edge deployments through the Pulsar engine’s two‑tier architecture, which separates compute from storage. A single core requires only 2 GB of RAM, and deployments can use local storage and buffering when cloud connectivity is intermittent. StreamNative Private Cloud can run on bare metal, Docker, Kubernetes, and KubeEdge (an open source CNCF project that extends Kubernetes to edge devices). Edge ingestion is supported via Pulsar and Kafka for gateway devices and MQTT for IoT devices. Local processing is handled by Pulsar Functions, which can aggregate, transform, and enrich data before syncing to the cloud. Pulsar Functions can also run ML model inference using embedded models supported by ONNX Runtime and TensorFlow Lite. Additional libraries supported by Pulsar Functions include Scikit-Learn, PyTorch and TensorFlow frameworks, and gradient boosting libraries like XGBoost, LightGBM, and CatBoost.

• SQL friendliness: StreamNative provides multiple SQL‑based options for working with streaming data. pfSQL is the company’s SQL‑like framework for message routing and filtering within Pulsar Functions. Through Flink SQL, organizations can process, transform, and query data in motion. The Flink Table API with the Pulsar SQL Connector enables SQL queries directly on Pulsar topics. For data at rest, Ursa enables SQL access to lakehouse tables in Delta Lake and Iceberg formats, which can be queried through engines such as Snowflake and Trino.

Opportunities

StreamNative has room for improvement in a couple of decision criteria, including:

• Streaming data visualizations: The platform includes a cloud console dashboard for monitoring resource utilization and cluster health, and it integrates with Grafana and Prometheus for observability and custom dashboards. Adding more native capabilities for visual data exploration and point‑and‑click pipeline building would strengthen StreamNative’s showing in this category.

• Advanced analytics: StreamNative underlies several facets of advanced analytics, including computer vision, RAG workflows, and various facets of vector embeddings and multiagent architectures. Expanding the system’s native geospatial analytics would likely be well received by certain customers.

Purchase Considerations

Private cloud deployments require an annual subscription based on deployment size. BYOC deployments incur cloud infrastructure costs plus a StreamNative management fee. Dedicated deployments use a monthly subscription model based on configuration and cluster size. The serverless deployment option uses a pay‑per‑use model in which compute and storage consumption determine cost.

Use Cases

StreamNative supports a variety of streaming applications, including IoT telemetry and sensor data processing, streaming to Delta Lake and Iceberg tables for unified historical and real‑time analytics, and AI agent orchestration for RAG workflows.

Striim

Solution Overview

Striim provides a distributed in‑memory streaming and data integration platform that supports CDC, streaming SQL processing, and in-flight transformations. Supported streaming engines include Amazon Kinesis, Apache Kafka, Google Pub/Sub, and Azure Event Hubs. Striim Cloud is accessible through AWS, Google Cloud, and Microsoft Azure. The platform can persist its messaging and queuing layers in Kafka to decouple publishers from subscribers, enabling organizations to process events in memory and route them to multiple downstream targets.

In 2025, Striim released Validata, a data reconciliation and validation product that works with numerous data replication and data integration tools. It includes schema drift detection, automated repair scripts, and reporting capabilities. Validata incorporates Validata AI, an LLM‑powered agent, to help with these and other tasks. Additional agents support PII detection, vector embedding generation, anomaly detection, time series forecasting, and ML within streaming topologies.

Striim is positioned as a Leader and Fast Mover in the Innovation/Platform Play quadrant of the streaming data Radar chart.

Strengths

Striim scored well on a number of decision criteria, including:

• SQL friendliness: Striim offers extensive SQL capabilities, including SQL2Fabric‑X, which enables replication from SQL Server to Microsoft Fabric Data Warehouse, Microsoft Fabric SQL database, and Azure SQL Database. The Azure SQL Database implementation supports schema evolution, vector‑embedding enrichment, inline transformations, and CDC. Tungsten Query Language (TQL) supports aggregation, transformation, and filtering, and ML models can enhance TQL usage for time series forecasting and anomaly detection. Striim’s SQL‑based distributed processing layer can query streaming data across single or multiple streams and partition workloads across clusters.

• Complex event stream processing: Striim supports time‑based, session‑based, and batch‑based windows, which can be combined into hybrid windows. Additional techniques include pattern‑matching windows with delimiters and custom parsing syntax for custom window definitions. The platform provides functions and operators for these specialized use cases. The MCP AgentLink feature replicates data from more than 100 sources (including Salesforce and PostgreSQL) into MCP‑read zones to support AI agents. It enriches data with vectors, metadata, joins, and in‑flight PII masking. Striim also generates vector embeddings through Euclid, an AI agent, and two AI agents that support PII discovery and protection. With AI Insights, the platform integrates with models from Azure OpenAI and Vertex AI and can store vector embeddings in repositories such as Snowflake, MongoDB, Databricks, BigQuery, and Oracle.

• Edge applicability: Striim supports edge deployments through a lightweight agent that runs on edge devices. The agent collects streaming data and forwards it to a centralized Striim cluster. It can be deployed on bare metal or containerized environments.

Opportunities

Striim has room for improvement in a couple of decision criteria, including:

• Advanced data quality: Validata includes custom SQL-based validation checks, six validation methods covering field-level content mismatches and schema drift, and AI-assisted discrepancy reporting. Expanding Striim’s native support for context‑aware string evaluation would strengthen its data quality capabilities.

• Event‑driven autoscaling: Striim supports elastic, horizontally scalable pipeline execution with one-click autoscaling for distributed workloads. Adding mechanisms specifically optimized for event‑driven scaling, like in-house, out-the-box native support for Dapr, would improve its showing in this area.

Purchase Considerations

Striim offers three pricing tiers: Striim Cloud, Striim Platform, and Striim Developer. Striim Cloud is the fully managed service, offering a free trial and providing horizontally scalable, distributed pipelines with dedicated compute and network resources. Striim Platform is the self‑managed deployment option, including self‑managed monitoring and alerts, clustered deployments, and high availability. The Developer tier is free and includes serverless stream processing with SQL and automated pipelines. Validata is not bundled into the existing tiers. It is an independent, standalone data validation product that can be purchased.

Use Cases

Striim supports enterprise-grade end-to-end data pipelines for cloud and data lake adoption, application rearchitecting for data and platform modernization, streaming analytics for anomaly detection, fraud detection, threshold deviations, dynamic decisioning, smart alerting, and operationalizing AI agents and workflows. It also offers support for contemporary statistical AI use cases, including real‑time RAG, retrieval augmented conversation, agentic AI, event-based prompt injection/responses, PII detection, and data cleansing.

TIBCO: TIBCO Streaming

TIBCO Streaming processes in‑flight data in real time at high volumes. The platform provides a shared UI and a large library of adapters that convert data between TIBCO Streaming and external systems. These adapters are licensed separately based on source types and customer needs. The vendor offers more than 150 such adapters, covering a wide range of verticals and technologies. Core components of the solution include a streaming server, a control center for low‑latency operational intelligence across multiple data sources, and an in‑memory data mart.

TIBCO Streaming integrates with the broader TIBCO ecosystem, including the Spotfire BI platform. Users can build analytics applications using EventFlow, TIBCO Streaming’s visual programming language, within StreamBase Studio, its integrated development environment. StreamBase Operators provide predefined runtime actions such as querying, merging, and other common processing tasks.

TIBCO is positioned as a Leader and Outperformer in the Maturity/Platform Play quadrant of the streaming data Radar chart.

Strengths

TIBCO scored well on a number of decision criteria, including:

• Advanced data quality: TIBCO Streaming offers multiple mechanisms to enhance data quality. Using EventFlow in StreamBase Studio, users can build applications that manage, transform, validate, standardize, and cleanse data. The BSort operator buffers and sorts events by timestamp or other fields to correct slight out‑of‑order arrival. Native regular expression functions support pattern detection, transformation, and validation. Additional functions address common data quality tasks such as casing and value constraints. Developers can also break unstructured strings (such as log lines or address fields) into structured tokens for downstream processing.

• Streaming data visualizations: TIBCO StreamBase LiveView, the vendor’s streaming BI platform, provides configurable dashboards and graphical or tabular views of live data. LiveView, along with Spotfire, enables users to query streaming data for operational intelligence, monitor ingestion and connectivity, track event processing and system health, and assess data quality. StreamBase Studio and the associated BI tools offer point‑and‑click and drag‑and‑drop interfaces for navigating and interacting with these capabilities.

• SQL friendliness: LiveView Query Language (LiveQL), TIBCO’s SQL‑like language, supports snapshot and continuous queries against data organized into tables. TIBCO also provides connectors for Apache Kafka, enabling SQL‑oriented users to consume Kafka topic data through familiar tools such as ksqlDB.

TIBCO was deemed an Outperformer due to its progressive stance on adding support for a wide range of ML models, including those that are generative.

Opportunities

TIBCO has room for improvement in a couple of decision criteria, including:

• Advanced analytics: Although TIBCO Streaming integrates with multiple ML libraries and supports invoking external models on streaming data, these capabilities are not native to the platform. Adding built‑in operators for real‑time model scoring or statistical analysis would strengthen the platform’s performance in this criterion.

• Vibe coding data pipelines: TIBCO is evaluating natural language pipeline authoring, where users describe a pipeline and a model generates the underlying code. Introducing this capability would enhance the platform’s usability and broaden its appeal to teams seeking low‑code or AI‑assisted development.

Purchase Considerations

Pricing for TIBCO Streaming depends on support level and whether the deployment is for production or nonproduction use. Licensing is subscription based, with sizing determined by the number and type of nodes and processors required. There are licenses required for all of the streaming adapters, which include Standard Adapters, Enterprise Streaming Premium Adapters, and Enterprise Streaming FIX Adapters. The ProdPlus license allows organizations to deploy the solution in their own chosen compute environment.

Use Cases

Supported use cases include operational intelligence, real‑time analytics, and data science scenarios such as scoring ML models for business applications.

VAST Data: VAST AI Operating System

Solution Overview

The VAST AI Operating System unifies streaming data ingestion, analytical queries, real‑time event processing, unstructured data storage, and transactional persistence in a single platform. Its primary components include VAST DataBase, VAST DataStore, VAST Event Broker, VAST DataEngine, and VAST DataSpace.

VAST DataBase is an analytical and transactional database that ingests streaming data and stores traditional structured data, including streaming data, into tables and vector embeddings. VAST DataStore is a flash‑optimized unstructured data store that exposes an S3‑compatible API. VAST Event Broker is a Kafka‑compatible messaging service that lands event streams into a Write Buffer backed by SCM (storage‑class memory). The broker provides transactional durability so data can be transformed, processed, materialized, or indexed for analytics in VAST DataBase. VAST DataEngine is the platform’s compute fabric, implementing serverless functions triggered by VAST Event Broker for metadata and object events. VAST DataSpace is a global namespace and data coordination layer connecting all VAST clusters across cloud, on‑prem, and edge environments. VAST can store and read multiple types of storage, including S3, block storage, NFS, vector embeddings, and structured tables.

VAST Data is positioned as a Challenger and Outperformer in the Maturity/Feature Play quadrant of the streaming data Radar chart.

Strengths

VAST Data scored well on a number of decision criteria, including:

• Advanced analytics: Users can build and deploy ML and AI pipelines with VAST Python Functions, which act as UDFs for integrating external libraries, data science workflows, and generative model resources. VAST Python Functions allow users to select embedding models and generate vector embeddings stored in VAST DataBase’s native vector store. Organizations can apply windowing logic to data once it is materialized as tables via VAST Event Broker. External engines such as Spark or Trino can be used to apply SQL‑based windowing, aggregations, and filtering.

• Complex event stream processing: VAST DataEngine uses VAST Python Functions to create user‑defined, event‑driven pipelines for training, scoring, and running inference jobs on ML and predictive models with streaming data. Users can employ their models of choice, and the platform supports ASTORE and ONNX formats, both of which allow customer‑specified logic. Once data streams are ingested into VAST Event Broker and materialized as tables, it is possible to perform joins between streaming sources and between streaming and historical data.

• SQL friendliness: VAST Event Broker stores topics as tables in VAST DataBase, which can be queried with SQL. VAST provides managed Trino and Spark engines running on VAST CNodes, and APIs allow connectivity to Dremio and other SQL engines. These engines enable organizations to perform complex joins, aggregations, and windowing on historical and streaming data stored in VAST.

VAST Data was classified as an Outperformer due to its aggressive roadmap strategy for agent-based workflows relying on streaming data to work with VAST DataStore and VAST DataBase.

Opportunities

VAST Data has room for improvement in a couple of decision criteria, including:

• Advanced data quality: Event triggers from VAST Event Broker can invoke serverless Python Functions for data quality tasks. Expanding these triggers beyond object and metadata events to include message‑level events would strengthen VAST Data’s data quality capabilities.

• Streaming data visualizations: VAST Management Service provides dashboards for monitoring streaming events and consumption. Enhancing these capabilities with support for “visually” querying data via point‑and‑click and drag‑and‑drop interfaces would make the vendor more capable in this aspect of streaming data platforms.

Purchase Considerations

With VAST Data’s Gemini subscription model, customers purchase infrastructure from hardware partners and subscribe to VAST based on compute and usable capacity. Subscriptions include customer support, maintenance, and all software features, including VAST DataBase, VAST DataStore, and the rest of the VAST AI Operating System.

Use Cases

VAST Data is well suited for security event processing, fraud detection, RAG with streaming data inputs, financial trading, and IoT analytics.

Organizations evaluating streaming data platforms must recognize the dual forces shaping today’s market: vendor consolidation and expansion of scope. Consolidation is evident in hyperscalers that now offer multiple overlapping streaming services, as well as in strategic acquisitions such as IBM’s purchases of Confluent and StreamSets. At the same time, streaming platforms are expanding into adjacent areas of data management (including data privacy, regulatory compliance, metadata management, and data federation) and, in doing so, are beginning to define a new baseline for what contemporary data platforms are expected to offer.

Buyers should keep both dynamics in view: how many services a vendor provides and how comprehensively those services address the full lifecycle of streaming data.

The more significant shift, however, is the rapid integration of AI into event stream processing. For many organizations, streaming has become synonymous with powering intelligent agents, RAG, and interactive AI applications. While these are among the fastest‑growing use cases, it would be a mistake to evaluate platforms solely through the lens of GenAI. Long‑standing verticals (including manufacturing, supply chain, digital twins, finance, insurance, healthcare, and the broader IoT ecosystem) continue to depend on streaming data regardless of whether generative models are involved. The enterprise value of a platform cannot be reduced to its AI story alone, even as AI‑driven workloads become impossible to ignore.

A pragmatic purchasing process begins with industry alignment. Organizations should identify which platforms have meaningful customer footprints in their target verticals and which vendors can demonstrate production‑ready use cases that resemble the workloads the buyer intends to deploy. Proofs of concept and targeted demonstrations can help clarify fit and surface operational considerations that may not be obvious from feature lists alone.

From there, buyers should evaluate platforms through the enduring criteria that define streaming systems: latency, scalability, resiliency, governance, and operational simplicity. These fundamentals remain the backbone of successful deployments, whether the workload is a high‑volume IoT pipeline, a fraud‑detection system, or an AI‑augmented application.

To learn about related topics in this space, check out the following GigaOm Radar reports:

• GigaOm Radar for Data Lakehouses

• GigaOm Radar for Data Access Governance

• GigaOm Radar for Data Warehouses

*Vendors marked with an asterisk did not participate in our research process for the Radar report, and their capsules and scoring were compiled via desk research.

For more information about our research process for Radar reports, please visit our Methodology.

Andrew Brust has held developer, CTO, analyst, research director, and market strategist positions at organizations ranging from the City of New York and Cap Gemini to GigaOm and Datameer. He has worked with small, medium, and Fortune 1000 clients in numerous industries and with software companies ranging from small ISVs to large clients like Microsoft. The understanding of technology and the way customers use it that resulted from this experience makes his market and product analyses relevant, credible, and empathetic.

Andrew has tracked the Big Data and Analytics industry since its inception, as GigaOm’s Research Director and as ZDNet’s original blogger for Big Data and Analytics. Andrew co-chairs Visual Studio Live!, one of the nation’s longest-running developer conferences, and currently covers data and analytics for The New Stack and VentureBeat. As a seasoned technical author and speaker in the database field, Andrew understands today’s market in the context of its extensive enterprise underpinnings.

Jelani Harper has worked as an information technology editorial consultant and journalist for over 10 years. During that time he has helped myriad vendors and publications in the data management space strategize, develop, compose, and place content in a variety of outlets, spanning mainstream outfits such as Forbes, Tech Crunch, and VentureBeat to trade journals like The New Stack, TDWI and KMWorld. He has produced an assortment of technical content including white papers, solutions briefs, contract proposals, marketing materials, thought leadership articles, bylines, and blogs for clients specializing in nearly every facet of data management.

As such, Jelani has focused extensively on the numerous dimensions of cognitive computing, cloud computing, analytics, data governance, data engineering, and data integration. His work has enabled him to conduct a number of substantive interviews with both established and progressive vendors in the arenas of High-Performance Computing, semantic technologies, quantum computing, cyber security, the Internet of Things, blockchain, and more. Content under his own byline has been cited or paraphrased by organizations such as SAS, TAMR, and DAMA International, as well as publications like VentureBeat, IT Business Edge, and The Data Administration Newsletter. He’s spent the last couple of years working with Blue Badge Insights.

GigaOm provides technical, operational, and business advice for IT’s strategic digital enterprise and business initiatives. Enterprise business leaders, CIOs, and technology organizations partner with GigaOm for practical, actionable, strategic, and visionary advice for modernizing and transforming their business. GigaOm’s advice empowers enterprises to successfully compete in an increasingly complicated business atmosphere that requires a solid understanding of constantly changing customer demands.

GigaOm works directly with enterprises both inside and outside of the IT organization to apply proven research and methodologies designed to avoid pitfalls and roadblocks while balancing risk and innovation. Research methodologies include but are not limited to adoption and benchmarking surveys, use cases, interviews, ROI/TCO, market landscapes, strategic trends, and technical benchmarks. Our analysts possess 20+ years of experience advising a spectrum of clients from early adopters to mainstream enterprises.

GigaOm’s perspective is that of the unbiased enterprise practitioner. Through this perspective, GigaOm connects with engaged and loyal subscribers on a deep and meaningful level.

© Knowingly, Inc. 2026 "GigaOm Radar for Streaming Data Platforms" is a trademark of Knowingly, Inc. For permission to reproduce this report, please contact sales@gigaom.com.