--- title: PostgreSQL 19 gets a REPACK command, Databricks... description: PostgreSQL 19 is adding a native REPACK command that consolidates VACUUM FULL and CLUSTER into a single operation, shrinking bloated tables significantly, with... canonical: https://daily.dev/posts/postgresql-19-gets-a-repack-command-databricks-structured-streaming-hits-millisecond-latency-f6touuhea twitter:card: summary_large_image twitter:site: @dailydotdev og:type: website og:site_name: daily.dev og:title: PostgreSQL 19 gets a REPACK command, Databricks Structured Streaming hits millisecond latency | daily.dev og:description: PostgreSQL 19 is adding a native REPACK command that consolidates VACUUM FULL and CLUSTER into a single operation, shrinking bloated tables significantly, with... og:url: https://daily.dev/posts/postgresql-19-gets-a-repack-command-databricks-structured-streaming-hits-millisecond-latency-f6touuhea og:image: https://api.daily.dev/og/posts/f6tOUUhEA.png og:image:alt: PostgreSQL 19 gets a REPACK command, Databricks Structured Streaming hits millisecond latency og:image:width: 1200 og:image:height: 630 og:locale: en --- > ## Documentation Index > Fetch the complete documentation index at: https://daily.dev/llms.txt > Use this file to discover all available pages before exploring further. # PostgreSQL 19 gets a REPACK command, Databricks Structured Streaming hits millisecond latency **[Databases Digest](https://daily.dev/sources/databases_digest)** · 4 min read · 0 upvotes · 0 comments ## Summary PostgreSQL 19 is adding a native REPACK command that consolidates VACUUM FULL and CLUSTER into a single operation, shrinking bloated tables significantly, with a future CONCURRENTLY mode planned. Databricks has GA'd Real-Time Mode for Spark Structured Streaming on Runtime 16.4+, replacing microbatching with continuous flow and achieving sub-100ms P99 latency for early adopters like Coinbase and DraftKings. A detailed walkthrough covers implementing the Outbox Pattern in Go with PostgreSQL using pgx, FOR UPDATE SKIP LOCKED, and WAL-based logical replication. SpacetimeDB's 1000x Postgres performance claim is contextualized as a narrow benchmark comparing co-located in-memory architecture against a networked disk-based system. Additional items cover DynamoDB internals, text-to-SQL agent accuracy, Snowflake graph queries with recursive CTEs, and reactive streaming with Spring WebFlux. ## Content **TLDR:** PostgreSQL 19 is adding a native REPACK command that replaces the VACUUM FULL/CLUSTER combo for reclaiming table bloat. Databricks just GA'd Real-Time Mode for Spark Structured Streaming, with early adopters reporting sub-100ms P99 latency. There's also a solid writeup on the Outbox Pattern in Go and PostgreSQL worth reading if you're building event-driven systems. SpacetimeDB is making noise with 1000x Postgres performance claims that deserve some scrutiny. --- ## PostgreSQL 19 introduces REPACK PostgreSQL 19 is adding a built-in REPACK command that consolidates VACUUM FULL and CLUSTER into one cleaner operation. In the examples shown, an 8.5GB bloated table (after deleting 80% of rows) shrinks to ~1.6GB. It supports ordering by a specified index, includes a new `pg_stat_progress_repack` view, and accepts VERBOSE and ANALYZE options. The current implementation acquires the same locks as VACUUM FULL, but the design explicitly leaves room for a future CONCURRENTLY mode — which is the part that actually matters for production workloads. ## Databricks Real-Time Mode for Spark Structured Streaming is generally available Databricks GA'd Real-Time Mode (RTM) for Apache Spark Structured Streaming on Databricks Runtime 16.4+, enabled via a single config flag. RTM replaces microbatching with continuous data flow, and Databricks claims latency comparable to Flink in benchmarks. Coinbase, DraftKings, and MakeMyTrip are cited as early adopters reporting 80%+ latency reductions and sub-100ms P99s for fraud detection and ML feature pipelines. If the benchmark holds up in practice, this removes a real reason to run a separate Flink cluster alongside Spark. ## Outbox Pattern in Go and PostgreSQL A thorough implementation walkthrough covers the full Outbox Pattern using pgx and PostgreSQL: storing pending events in an outbox table within the same transaction as business data, then polling with a relay using `FOR UPDATE SKIP LOCKED` for concurrent-safe processing. The post covers at-least-once delivery semantics, idempotency strategies for consumers, and a higher-throughput alternative using PostgreSQL logical replication via the WAL. If you've been duct-taping this together yourself, it's worth comparing against. ## SpacetimeDB's 1000x Postgres claim SpacetimeDB is circulating benchmark results claiming over 1000x faster performance than PostgreSQL. The architectural explanation is straightforward: application code and database run on the same machine (no network round-trips), and queries run against in-memory data with disk persistence in the background — essentially Redis with a query engine. The 1000x number is real in the narrow scenario they're measuring, but it's comparing a co-located in-memory system against a networked disk-based one. Not a fair fight, and not a drop-in replacement. --- ## Also notable - **Operational ontologies in Snowflake:** A practical guide to storing hierarchical data (BOMs, supplier tiers) as node-edge tables and precomputing transitive relationships with closure tables built from recursive CTEs, turning multi-hop graph traversals into flat indexed SQL lookups that run under a second for 40,000+ component BOMs. - **Text-to-SQL agent analysis:** A 50-question BIRD-Bench sample run through Claude Opus 4.5 with MotherDuck MCP found single-shot answers succeed 91% of the time while iterative loops succeed only 64% — and agents that struggle tend to fail completely rather than partially. - **DynamoDB architecture deep dive:** Covers the full request path through five internal services (request router, auth, metadata service, global admission controller, storage nodes), quorum-based writes, WAL archival to S3, and the MEMSDS in-memory metadata cache that prevents thundering herd problems. - **KeeperDB launch:** Keeper Security added vault-embedded database access to KeeperPAM, supporting MySQL, PostgreSQL, Oracle, and SQL Server with session recording and a proxy option for existing database clients. Official launch at RSA Conference 2026. - **Amazon RDS Custom for SQL Server OS updates:** Customers can now view and schedule OS updates for RDS-provided engine versions via `describe-pending-maintenance-actions` and `apply-pending-maintenance-action` APIs, available across all supported regions. - **Snowflake graph reasoning without a graph database:** Closure tables built from recursive CTEs handle multi-hop relationship queries (part_of, suppliedBy, locatedIn) in flat SQL, with Cortex Analyst sitting on top for natural language querying. - **Reactive streaming with Project Reactor:** A step-by-step build of a non-blocking SSE endpoint using Spring WebFlux and Reactive MongoDB, with explicit backpressure via `onBackpressureLatest()` and a clear warning that calling `.block()` anywhere destroys the scalability model. - **SQL transformation discipline:** A case for treating dbt/SQLMesh transformation layers like software — modular, version-controlled, tested, with structured modeling layers (raw, staging, intermediate, marts) and no business logic buried in BI tools. - **PGConf.dev 2026:** Scheduled for May 19-22 in Vancouver, with community-led sessions added this year and events marking the 30th anniversary of the PostgreSQL project. --- Tags: [#backend](https://daily.dev/tags/backend), [#postgresql](https://daily.dev/tags/postgresql), [#apache-spark](https://daily.dev/tags/apache-spark), [#databricks](https://daily.dev/tags/databricks) [View this post on daily.dev](https://daily.dev/posts/postgresql-19-gets-a-repack-command-databricks-structured-streaming-hits-millisecond-latency-f6touuhea) ```json {"@context":"https://schema.org","@graph":[{"@type":"Organization","@id":"https://daily.dev/#organization","name":"daily.dev","url":"https://daily.dev","logo":{"@type":"ImageObject","url":"https://daily.dev/apple-touch-icon.png","width":180,"height":180},"sameAs":["https://twitter.com/dailydotdev","https://github.com/dailydotdev","https://www.linkedin.com/company/daily-dev-ltd"]},{"@type":"WebSite","@id":"https://daily.dev/#website","url":"https://daily.dev","name":"daily.dev","publisher":{"@id":"https://daily.dev/#organization"},"potentialAction":{"@type":"SearchAction","target":{"@type":"EntryPoint","urlTemplate":"https://daily.dev/search?q={search_term_string}"},"query-input":"required name=search_term_string"}}]} {"@context":"https://schema.org","@type":"DiscussionForumPosting","mainEntityOfPage":"https://daily.dev/posts/postgresql-19-gets-a-repack-command-databricks-structured-streaming-hits-millisecond-latency-f6touuhea","headline":"PostgreSQL 19 gets a REPACK command, Databricks Structured Streaming hits millisecond latency","text":"PostgreSQL 19 is adding a native REPACK command that consolidates VACUUM FULL and CLUSTER into a single operation, shrinking bloated tables significantly, with a future CONCURRENTLY mode planned. Databricks has GA'd Real-Time Mode for Spark Structured Streaming on Runtime 16.4+, replacing microbatching with continuous flow and achieving sub-100ms P99 latency for early adopters like Coinbase and DraftKings. A detailed walkthrough covers implementing the Outbox Pattern in Go with PostgreSQL using pgx, FOR UPDATE SKIP LOCKED, and WAL-based logical replication. SpacetimeDB's 1000x Postgres performance claim is contextualized as a narrow benchmark comparing co-located in-memory architecture against a networked disk-based system. Additional items cover DynamoDB internals, text-to-SQL agent accuracy, Snowflake graph queries with recursive CTEs, and reactive streaming with Spring WebFlux.","url":"https://daily.dev/posts/postgresql-19-gets-a-repack-command-databricks-structured-streaming-hits-millisecond-latency-f6touuhea","datePublished":"2026-03-20T04:17:48.615Z","dateModified":"2026-03-20T04:18:08.895Z","author":{"@type":"Organization","name":"Databases Digest","logo":"https://media.daily.dev/image/upload/s--ir1hdV8b--/f_auto,q_auto/v1773839400/logos/databases_digest?_a=BAMAMiiu0","url":"https://daily.dev/sources/databases_digest"},"interactionStatistic":[{"@type":"InteractionCounter","interactionType":{"@type":"LikeAction"},"userInteractionCount":0},{"@type":"InteractionCounter","interactionType":{"@type":"CommentAction"},"userInteractionCount":0}],"isPartOf":{"@type":"WebPage","url":"https://daily.dev/sources/databases_digest","name":"Databases Digest"}} {"@context":"https://schema.org","@type":"BreadcrumbList","itemListElement":[{"@type":"ListItem","position":1,"name":"Home","item":"https://daily.dev"},{"@type":"ListItem","position":2,"name":"Databases Digest","item":"https://daily.dev/sources/databases_digest"},{"@type":"ListItem","position":3,"name":"PostgreSQL 19 gets a REPACK command, Databricks Structured Streaming hits millisecond latency"}]} ```