AWS Glue 6.0 Launch: 30% Lower Cost, Full Iceberg v3 Support Now

AWS unveiled Glue version 6.0, slashing the service’s price by roughly a third while adding native support for Apache Iceberg v3. The upgrade runs on Spark 4.1, Python 3.13 and Scala 2.13, giving data engineers a faster, more flexible runtime for lake‑house pipelines. For enterprises that already rely on AWS for ETL, the price cut and full Iceberg compatibility translate into immediate savings and smoother migration to modern data‑mesh architectures, a shift that could reshape budgeting decisions across cloud‑first organizations today.

Glue 6.0 replaces the legacy Spark 3.x engine with Spark 4.1, a release that brings adaptive query execution, better memory management and accelerated streaming support. The runtime now ships with Python 3.13, offering newer language features and performance improvements, while Scala 2.13 reduces binary incompatibilities for existing libraries. Crucially, the service embeds the full Iceberg v3 catalog, enabling ACID transactions, hidden partitioning and time‑travel queries without external glue‑catalog glue‑jobs. Pricing is now measured per DPU‑hour, with a 30 % reduction compared to Glue 3.0, making large‑scale batch jobs markedly cheaper.

Across the cloud data‑integration market, cost‑efficiency and lake‑house readiness have become decisive factors. Competitors such as Databricks and Snowflake have been touting unified analytics platforms that natively support Iceberg or Delta Lake. Google Cloud’s Data Fusion also recently added Iceberg connectors, but its pricing remains higher for comparable workloads. Industry analysts note that the global managed‑ETL market, valued at $7 billion in 2023, is projected to grow 18 % YoY, driven by enterprises consolidating data pipelines onto single‑cloud services that promise lower operational overhead.

In India’s rapidly expanding data ecosystem, the Glue 6.0 price cut could be a game‑changer for fintech firms, e‑commerce platforms and media streaming services that process petabytes of click‑stream and transaction data. Companies like Razorpay, Swiggy and Byju’s already run massive Spark jobs on AWS; a 30 % reduction in compute spend could free millions of rupees for AI model training or edge‑analytics initiatives. Moreover, the native Iceberg support aligns with the Indian government’s push for data‑sovereignty, as it simplifies building compliant lake‑house solutions that retain granular audit trails.

Key Highlights

• Launches Glue 6.0 with Spark 4.1, Python 3.13 and Scala 2.13 runtime
• Adds full Apache Iceberg v3 support for ACID transactions and time‑travel queries
• Reduces Glue compute pricing by roughly 30 % per DPU‑hour
• Targets data engineers and ETL developers seeking lower‑cost lake‑house pipelines
• Future roadmap includes serverless Spark 5.x and deeper integration with AWS Lake Formation

Real-World Impact

Data engineers can now rewrite existing Glue jobs to leverage Spark 4.1’s optimizer, cutting job duration by up to 20 %. Financial analysts will see lower cloud‑cost line items, while startup CTOs can allocate saved budgets toward real‑time analytics or machine‑learning workloads. The change also benefits AWS partners who build managed data‑platform services, giving them a more competitive price point against rival cloud providers.

Why This Matters

The update signals AWS’s commitment to the lake‑house paradigm, positioning Glue as a cost‑effective alternative to proprietary data‑warehousing services. For CTOs, the reduced spend and native Iceberg compatibility mean they can consolidate ETL, catalog and governance layers onto a single AWS stack, simplifying architecture and accelerating time‑to‑insight. Developers should start testing Iceberg tables in staging environments to exploit transaction guarantees before migrating production pipelines.

As cloud providers race to lock in the lake‑house market, Glue 6.0’s price advantage and full Iceberg support set a new benchmark. Watching how AWS rolls out serverless Spark 5.x later this year will indicate whether the cost lead can be sustained against emerging multi‑cloud data‑fabric solutions.