Classmethod Data Analytics Newsletter (AWS Data Analytics Edition) – October 2026 Issue
This page has been translated by machine translation. View original
I'm Ishikawa from the Consulting Department, Cloud Business Division. Here are the AWS data analytics updates for September 2026. This month, Aurora PostgreSQL can now query Apache Iceberg and Parquet data directly with DuckDB without ETL, Amazon EMR introduced a Long Term Support (LTS) release based on Apache Spark 4.1, and Amazon S3 Tables now supports all Apache Iceberg V3 data types. Iceberg-related updates were announced across multiple services: Aurora, EMR, S3 Tables, and SageMaker Unified Studio. There are other updates too, so let me introduce them!
Amazon Redshift
New features and updates
2026/09/03 - Amazon Redshift rg.large instances now support single-node clusters
You can now create single-node clusters with Amazon Redshift rg.large instances powered by AWS Graviton processors. This is intended for small workloads that don't require high availability, and lets you run proofs of concept and tests quickly at low cost. It is available with patch version P204 and later. RG instances deliver up to 2.4 times the performance of the previous-generation RA3 instances at a 30% lower price per vCPU. It is available in regions including the Tokyo and Osaka regions.
AWS Glue
New features and updates
2026/09/14 - AWS Glue zero-ETL adds target table property ownership and conflict detection
AWS Glue zero-ETL integrations now record which integration owns the properties of a target table and detect conflicts. Two integrations will no longer end up pointing at the same target table without anyone noticing. If you try to create or modify an integration that uses table properties owned by another integration, Glue shows the owning integration and guides you to choose a different target or update the existing integration. It applies to Amazon S3 Tables and SageMaker Lakehouse catalogs, and is available in all commercial regions and the GovCloud (US) regions that support zero-ETL integrations.
2026/09/15 - AWS Glue zero-ETL now captures archived Salesforce records
AWS Glue zero-ETL integrations with Salesforce as the source now bring archived records into the target table as well. The archive status is reflected in the isArchived field. Large entities that Salesforce archives automatically, such as Events, Tasks, and Activities, can now also be included in analysis. New integrations capture them automatically from creation, and existing integrations automatically backfill the archive status without resyncing or schema changes. It is available in all commercial regions and the GovCloud (US) regions that support Salesforce zero-ETL integrations.
2026/09/22 - AWS Glue Data Quality delivers context-specific rule recommendations in seconds
An Advanced mode for rule recommendations has been added to AWS Glue Data Quality. It uses generative AI to determine how the data is used and proposes data quality rules covering all columns of a table in the AWS Glue Data Catalog within seconds. You can prepare initial rules for newly ingested datasets and basic checks across a large data lake without writing rules one by one. You can review and adjust the proposed rules, save them as a ruleset, and start monitoring right away. It is available in 17 regions, including the Tokyo and Osaka regions.
API changes
2026/09/14 - AWS Glue - 1 new 3 updated methods
A new API, ListIntegrationTableProperties, has been added, and IntegrationArn has been added to TargetTableConfig.
2026/09/18 - AWS Glue - 3 updated methods
An Advanced mode has been added to AWS Glue Data Quality rule recommendations. It analyzes samples of table data with Amazon Athena and recommends DQDL rules with Amazon Bedrock.
2026/09/22 - AWS Glue - 8 updated methods
Two fields, SubObjectsStatistics and SparkPipelineInfo, have been added to the Glue materialized view feature.
2026/09/25 - AWS Glue - 3 updated methods
Table-level federation is now supported.
2026/09/29 - AWS Glue - 8 updated methods
Glue system-managed materialized views are now supported.
2026/09/30 - AWS Glue - 11 updated methods
You can now specify a Catalog ID for crawlers, column statistics, and materialized views.
Amazon Quick
New features and updates
2026/09/01 - Amazon Quick now lets you build custom apps with natural language
The feature in Amazon Quick that lets you create custom applications simply by describing the application you want in natural language is now generally available. You can create project trackers, customer-facing dashboards, training portals, and more without writing code. Applications connect to existing data such as Salesforce, Jira, ServiceNow, Microsoft 365, databases, and data warehouses, and update automatically as the source data changes. They follow existing identity, authorization, and access control policies, and can be shared with specific users or the entire organization. It is available to Plus, Professional, and Enterprise users.
2026/09/02 - Amazon Quick adds new tool settings and Model Context Protocol (MCP) sync support for connectors
Tool settings and MCP sync have been added to Amazon Quick connectors. Connector owners and administrators can enable or disable individual tools within a connector, so that only approved tools are provided to end users. For each tool, you can also set whether consent is required before execution or whether it is left to the end user's discretion. With MCP sync, when tools are added or descriptions are updated on an external MCP server, the changes are automatically reflected in the connector. It is available in all regions where Amazon Quick is available.
2026/09/03 - Introducing Amazon Quick Max: 5x the usage for power users who want the most out of Quick
A new plan, Quick Max, has been added to Amazon Quick. It offers 5 times the usage and 5 times the storage of the Plus plan, so you can run more agents and workflows at scale and concurrently throughout the month. You can choose monthly or annual billing. From the Plus plan, you switch by selecting "Upgrade plan" from your name at the bottom of the left navigation bar.
2026/09/09 - Amazon Quick desktop app is now generally available on macOS and Windows
The Amazon Quick desktop app is now generally available on macOS and Windows. It can work directly with local files, and connects in the background to calendars, email, and business applications. Conversations, context, and agents are synced between the desktop and mobile apps. Quick agents also continue running in the background after you close your computer. It is available in 7 regions, including the Tokyo region.
2026/09/09 - Amazon Quick activity feed now available on iOS and Android mobile devices
The Amazon Quick activity feed is now available in the iOS and Android mobile apps. The activity feed summarizes items that need action across internal communications and tools in priority order, and lets you respond directly from Quick. Users decide which notifications are important, and Quick learns priorities through usage. You can also set up routines that post information from connected data sources to the feed. The mobile app is included in existing Amazon Quick entitlements at no additional charge.
2026/09/09 - Amazon Quick adds always-on agents, a sharper feed, and enterprise controls
Features for organizational management and answer verification have been added to Amazon Quick. Scheduled tasks and monitoring agents run in the cloud, and results reach the feed even when your laptop is closed. The feed gains top-level filters, a daily briefing updated three times a day, and search across up to 7 days of data.
For administrators, mobile device management, per-user custom permissions, and integration with Microsoft Purview data loss prevention have been added. Answers now include inline citations showing their sources.
Monthly agent hours have also increased from 4 to 8 for Professional and from 8 to 18 for Enterprise.
2026/09/17 - Amazon Quick adds Generate Sheet and image-based analysis creation
Two ways to build dashboards faster have been added to Generate Analysis in Amazon Quick. With Generate Sheet, when you describe the sheet you want in natural language, a sheet with visuals suited to your data, filter controls, and calculated fields such as year-over-year and month-over-month is added to an existing analysis. With the feature to generate an analysis from an image, attaching an image of a dashboard, including one from another BI tool, recreates it as an editable analysis.
At launch it is available to Author Pro users on the Enterprise subscription, and Authors can also use it as a promotion until December 2026.
API changes
2026/09/22 - Amazon QuickSight - 4 updated methods
Fine-grained custom permissions are now supported for 28 action connectors, including Gmail, Google Drive, Google Sheets, Airtable, and Dropbox. Administrators can allow or deny each connector operation, rather than the action connector as a whole.
Amazon SageMaker Unified Studio
New features and updates
2026/09/02 - Amazon SageMaker Unified Studio CI/CD adds notebook promotion and AI-assisted manifest generation
Two features have been added to the open-source deployment toolkit for CI/CD in Amazon SageMaker Unified Studio. The first is the agent skill generate-bundle-manifest, which inspects a project's connections, storage, and workflows to generate a deployment manifest. It applies least-privilege IAM guidance and replaces hard-coded resource IDs with environment variables.
The second is a feature to promote SageMaker Unified Studio notebooks between environments. The first deployment creates the notebook, and subsequent deployments update it while keeping the execution history. It is available in all regions where SageMaker Unified Studio is offered.
2026/09/03 - Amazon SageMaker Unified Studio Workflows support Python and Bash operators
PythonOperator and BashOperator can now be used in Amazon SageMaker Unified Studio workflows. In serverless workflows, you can run your own Python functions and shell commands directly without provisioning separate compute resources. You no longer need to split processing out to Lambda or ECS for data transformations or running shell scripts. In the task panel of a serverless visual workflow, you search for the operator, place it on the canvas, and specify the function or command in a script file. It is available in all regions where SageMaker Unified Studio is available.
2026/09/11 - Amazon SageMaker Unified Studio now supports ODBC connections for Microsoft Power BI and other analytics tools
Amazon SageMaker Unified Studio now supports ODBC connections, so ODBC-compatible tools such as Microsoft Power BI can connect directly to data under project governance using the Amazon Athena ODBC driver. The project overview page now shows ODBC connection information in addition to the existing JDBC. With Athena ODBC driver 2.2.0.1 or later, you can choose between a mode that signs in from the browser with IAM Identity Center and an external IdP, and a mode that uses temporary AWS credentials. It is available at no additional charge in all regions where SageMaker Unified Studio is available.
2026/09/28 - Amazon SageMaker Unified Studio now supports two new connection capabilities: Iceberg REST Catalog connections and IAM authentication for Amazon DocumentDB
Amazon SageMaker Unified Studio now supports Iceberg REST Catalog (IRC) connections and IAM authentication for Amazon DocumentDB. IRC connections can connect to Snowflake Open Catalog (Polaris), Databricks Unity Catalog, and generic IRC, and can be used for browsing in Data Explorer, reading and writing in Visual ETL, and querying from data notebooks.
With IAM authentication for DocumentDB, you authenticate with the connection's IAM role, without storing a username and password in connections or notebooks. It applies to instance-based clusters with DocumentDB 5.0 or later and TLS enabled. It is available at no additional charge in all regions where SageMaker Unified Studio is available.
Amazon DataZone
New features and updates
API changes
2026/09/18 - Amazon DataZone - 4 updated methods
A parameter for specifying the notebook type has been added.
2026/09/24 - Amazon DataZone - 5 updated methods
CreateEnvironmentBlueprint, UpdateEnvironmentBlueprint, GetEnvironmentBlueprint, and ListEnvironmentBlueprints now support TOOLING, a blueprint category for custom tools. Also, roleArn can now be specified in iamProperties of CreateConnection.
2026/09/30 - Amazon DataZone - 2 updated methods
You can now specify notebook run notification configurations.
Amazon EMR
New features and updates
2026/09/08 - Amazon EMR extends support at no additional cost for customers migrating off older releases
Workloads migrating off older Amazon EMR releases can now continue to receive support after end of support at no additional cost. Releases 5.36 and 6.6 to 6.15 will receive critical security fixes on a best-effort basis until June 30, 2027, and Standard Support for 7.0 to 7.10 is extended until August 31, 2027. To receive the extension, contact AWS Support with your migration plan. All deployment models are covered: EMR on EC2, EMR on EKS, and EMR Serverless. It applies in all regions where Amazon EMR is available.
2026/09/22 - Amazon EMR introduces Long Term Support with Apache Spark 4.1
Long Term Support (LTS) releases have been introduced for Amazon EMR. The first LTS release is emr-spark-8.1.0 (Apache Spark 4.1), which provides fixes for security, bug, and data corruption issues of critical and high severity for 36 months at no additional charge.
This release fully supports Apache Iceberg v3, so you can use geospatial, high-precision timestamp, and schema evolution capabilities. From Spark SQL, you can reference cross-account and Amazon S3 Tables catalogs by name, and the Iceberg, Delta Lake, and Apache Hudi table formats are detected automatically. Fine-grained access control has also been extended to more Iceberg operations and to Delta Lake VACUUM.
2026/09/22 - Amazon EMR 7.14 is now available
Amazon EMR 7.14 has been released, upgrading Apache Spark to 3.5.8 and Apache Iceberg to 1.10.1. Iceberg materialized views on tables with updates or deletes now refresh faster because change data capture reads only the data files that changed. EMR on EKS supports Spark Connect endpoints and IPv6 EKS clusters, and EMR Serverless raises the storage limit for Spark jobs from 200 GiB to 1 TiB. It is available in all regions where Amazon EMR is available.
2026/09/23 - Amazon EMR on EKS now supports IPv6 Amazon EKS clusters
Amazon EMR on EKS can now run workloads on IPv6 Amazon EKS clusters. Because you can use the IPv6 address space, settings to conserve IPv4 addresses such as secondary CIDRs and prefix delegation are no longer needed, and large jobs such as a Spark job with 500 executors can run concurrently without worrying about address limits. You can submit jobs from StartJobRun, Spark Connect interactive endpoints, and SageMaker Unified Studio without additional configuration. It is available at no additional charge with emr-7.14.0 and emr-spark-8.0.0 and later.
2026/09/24 - Run interactive workloads on Amazon EMR on EKS with Spark Connect
Amazon EMR on EKS now supports interactive Apache Spark sessions with Spark Connect. From SageMaker Unified Studio managed notebooks or IDEs such as Jupyter and Visual Studio Code, you can connect to Spark on an existing Amazon EKS cluster and develop and debug applications interactively. Sessions keep the Spark context across cells and scripts, and each session is protected by an IAM execution role. It is available in all commercial regions with EMR 7.14 (Spark 3.5) and emr-spark-8.1.0 (Spark 4.1).
Amazon Kinesis
New features and updates
2026/09/01 - Amazon Kinesis Data Streams now supports a dry run feature to validate API requests
Amazon Kinesis Data Streams now has a dry run feature that lets you check whether an API request would succeed without executing it. When you set the DryRun parameter to true in a request, a DryRunOperationException is returned if all checks succeed. Previously, you had to resort to methods such as deliberately sending an oversized PutRecord to check permissions, and if the limit changed, records could be written to the production stream. It covers five APIs: PutRecord, PutRecords, GetRecords, GetShardIterator, and SubscribeToShard, and is available in all regions where Kinesis Data Streams is available.
2026/09/23 - Amazon Kinesis Data Streams announces Service-Managed Partition Keys for simplified data ingestion
Service-managed partition keys are now available in the on-demand modes of Amazon Kinesis Data Streams (On-Demand Standard and On-Demand Advantage). When you send data without specifying a partition key, records are automatically distributed across shards according to the available warm capacity. For workloads where record ordering is not needed, such as log aggregation, metric collection, and IoT telemetry, you can avoid the skew across shards and throttling caused by random keys such as UUIDs. It can be used by upgrading to the latest AWS SDK or KPL, and is available at no additional charge in all commercial regions.
API changes
2026/09/01 - Amazon Kinesis - 5 updated methods
Dry runs that validate permissions and request parameters are now supported in the data plane APIs. If all checks succeed, a DryRunOperationException is returned.
2026/09/23 - Amazon Kinesis - 1 new 2 updated methods
On-demand streams now support a mode in which the service manages record distribution. When the distribution strategy is set to AUTO, records are distributed evenly across shards. You can set it at creation with CreateStream, or change it later with UpdateStreamRecordDistributionStrategy.
Amazon MSK
New features and updates
2026/09/29 - AWS Transform now supports Apache Kafka migration assessments for Amazon MSK
AWS Transform's agentic migration assessment can now evaluate migrations from on-premises Apache Kafka clusters to Amazon MSK. When you upload a Kafka cluster inventory file or describe the cluster configuration in chat, it checks compatibility with MSK in terms of topology, version, configuration, authentication, and quotas, recommends MSK Express broker sizes, and estimates costs for broker hours, storage, and data transfer. You can also create estimates with different region and retention period conditions. It is available in all regions where AWS Transform is offered.
API changes
Amazon MWAA
New features and updates
2026/09/01 - Amazon MWAA supports Apache Airflow version 3.3.1
Apache Airflow 3.3.1 is now available on Amazon MWAA. Airflow 3.3 adds a Task and Asset State Store that lets tasks retain state across retries and reruns, and a Language Task SDK (experimental feature) that lets you write task logic in Java or Go. It also adds expanded asset partitioning, pluggable retry policies, and bulk operations on DAG runs and task instances. In addition to creating new environments, it supports upgrades from 3.2 and later, and is available in all regions supported by MWAA.
2026/09/03 - Amazon MWAA adds built-in monitoring with Amazon CloudWatch
Monitoring using Amazon CloudWatch metrics has been added to the environment details page in the Amazon MWAA management console. The dashboard of key metrics can overlay recommended warning ranges. It also shows a list of CloudWatch alarms associated with the environment, and lets you create recommended alarms in bulk from AWS-managed templates. It applies to MWAA provisioned environments and is available in all regions where MWAA is available. Standard CloudWatch pricing applies to metric queries and alarms.
2026/09/04 - Amazon MWAA Serverless is now available in AWS GovCloud (US)
Amazon MWAA Serverless is now available in the AWS GovCloud (US-East) and AWS GovCloud (US-West) regions. MWAA Serverless is a service that automatically provisions and scales Apache Airflow compute resources on demand, and charges only for workflow execution time.
API changes
Amazon OpenSearch Serverless
New features and updates
2026/09/10 - Amazon OpenSearch Serverless is now available on v0 by Vercel
Amazon OpenSearch Serverless can now be used on v0, Vercel's platform for creating production web applications from ideas with AI. With natural language prompts, you can create full-stack applications that use full-text search and vector search for RAG. v0 generates the application code and even handles creating the OpenSearch Serverless collection, indexing data, and setting environment variables. It is available in 17 regions, including the Tokyo and Osaka regions.
Amazon OpenSearch Service
New features and updates
API changes
2026/09/29 - Amazon OpenSearch Service - 3 updated methods
In domain configuration changes, non-critical pre-checks are now shown as warnings without stopping processing, and you can acknowledge them and proceed with the new AcceptedWarnings parameter. Severity is shown in the change progress and dry run results.
Amazon S3 Tables
New features and updates
2026/09/30 - Amazon S3 Tables now support all Apache Iceberg V3 data types
Amazon S3 Tables now supports the geometry, geography, unknown, and nanosecond-precision timestamp data types from the Apache Iceberg V3 specification, as well as column default values. You can store geospatial coordinates and nanosecond-precision event times as they are, without converting them to strings or integers. Together with the already supported Variant type, deletion vectors, and row lineage, all data types introduced in V3 are now supported. With column default values, existing rows get a value for a newly added column without backfilling. It is available in all regions where S3 Tables is available.
Amazon S3 Vectors
New features and updates
2026/09/30 - Amazon S3 Vectors introduces metadata pre-filtering for up to 5x higher recall on filtered search
Amazon S3 Vectors now supports pre-filtering, which evaluates metadata filters before the similarity search. When the filter conditions are restrictive, it can return up to 5 times more vectors matching the conditions. The $startsWith operator, which filters values such as paths and URLs by prefix match, has also been added. Indexes in new vector buckets use pre-filtering by default, and existing indexes can be switched with the UpdateIndexMode API. It is available at no additional charge in all commercial regions and the China regions where S3 Vectors is available.
API changes
2026/09/30 - Amazon S3 Vectors - 2 new 3 updated methods
Metadata pre-filtering is now supported, improving recall for queries that specify filters.
Other notable updates
2026/09/30 - Aurora PostgreSQL now supports querying of Apache Iceberg and Parquet data
Aurora PostgreSQL can now directly query Apache Iceberg and Parquet data in Amazon S3, Amazon S3 Tables, and the AWS Glue Data Catalog without ETL pipelines or data replication. When you create an external table that references this data, queries are executed by the DuckDB query engine built into PostgreSQL. Through the Glue Data Catalog, you can also reference external Iceberg REST Catalog-compatible catalogs.
For latency-sensitive workloads, you can load data into Aurora native tables with SQL. It is available at no additional charge in all commercial regions and the GovCloud (US) regions with Aurora PostgreSQL 17.11, 18.6, and later.
Conclusion
In September 2026, updates related to Apache Iceberg were announced across multiple services. The Amazon EMR LTS release emr-spark-8.1.0 fully supports Iceberg v3, and Amazon S3 Tables now supports all V3 data types. SageMaker Unified Studio can connect to external Iceberg REST Catalogs, and Aurora PostgreSQL can now query Iceberg and Parquet data directly. Amazon Quick had 7 updates, including app creation with natural language and the general availability of the desktop app. If any updates catch your eye, please give them a try. I hope this article helps someone.
