Cloud & DevOpsNote
What the 2024 pipeline design on this blog costs at AWS's published Glue rates, why timestamp sync misses deletes, and what AWS changed, as of September 2026.
The pipeline this blog described in 2024, AWS Glue jobs copying changed rows into Redshift every hour, costs about US$1,270 a month per job at AWS’s list rate, and by design it never sees a row that was deleted. Our 2024 post reported that the design, run hourly at data-backup providers, moved 60 million records (45 GB) in 40 minutes. That is still a sound way to end data silos between microservices. Two things have changed since: what the law in India now asks of the copy in the warehouse, and what AWS itself now offers.
AWS bills Glue’s Spark jobs at US$0.44 per DPU-hour, by the second. Its own pricing example uses a six-DPU job and prices the same job on Flex execution, meant for work that can wait for spare capacity, at a rate that works out to US$0.29.1
One hourly 40-minute Glue job lists at about US$1,267 a month; Flex cuts it by a third
Monthly list cost in US$ of one six-DPU job running 40 minutes every hour, 720 runs, at AWS's published rates
Source: AWS Glue pricing page; TechBag arithmetic · as of 30 September 2026 · Redshift compute and storage are billed separately in every option
The 2024 design synced rows whose created_timestamp or updated_timestamp had moved since the last run. A deleted row has no timestamp to move, so the warehouse keeps it; the original post said as much, and pointed to soft-delete columns or triggers. India’s Digital Personal Data Protection Act, 2023 makes that gap a compliance question. Section 8(7) requires a data fiduciary to erase personal data, and have its processors erase it, once consent is withdrawn or the purpose is served; section 12 gives the person a right to ask.2
One of the four AWS services our 2024 post named had already closed to new customers: AWS Data Pipeline stopped taking them on 25 July 2024, 48 days before that post, and AWS points migrating users to Glue, Step Functions or Managed Workflows for Apache Airflow.3 It is in maintenance mode, with no new features planned.4 For sources in Aurora, AWS now offers zero-ETL integration with Redshift at no additional fee, with data available “within seconds”; the change-data-capture I/O, the seed export and Redshift’s storage and compute are still billed.5 Because it works from captured changes rather than timestamps, it is the first design to test for deletes: delete a test row and check the warehouse.
Buying it in India. AWS publishes Glue in US dollars and says pricing can vary by Region, so price the job in the region the data sits in, Mumbai or Hyderabad.1 An Indian invoice adds 18% GST. At ₹83 to the dollar, the standard-execution job above is about ₹1,05,178 a month, or ₹1,24,110 with GST. The DPDP Rules, notified on 13 November 2025, bring the erasure and security duties into force on 13 May 2027.6 The Act’s schedule sets penalties of up to ₹250 crore for failing to keep reasonable security safeguards.2 Our databases and data tools guide covers the engines these pipelines feed.
Sources
Disclosure. TechBag is a software reseller: it earns a margin when a reader buys through it, including from vendors named here. The analysis is TechBag’s own; no vendor paid for or saw this piece before publication. Facts are labelled reported, TechBag count or our view.