We wrote up more about TanStack DB with ElectricSQL here: neon.com/blog/tanstac...

TanStack DB adds optimistic updates on top, making your UIs feel snappy, not just for GET requests, but also when mutating data from the UI.

ElectricSQL moves state synchronization into a dedicated service, greatly simplifying backend complexity. Instead of managing WebSockets, Server-Sent Events, or polling connections yourself, the ElectricSQL server handles the client updates for you.

Whenever an agent, background task, or webhook function mutates the database, the changes instantly propagate to all connected client devices, and the associated UIs update in real time.

This is fantastic not only for multiplayer or multi-device flows, but also when working with background AI agents and third-party services like Stripe triggering webhooks.

It's magical when you update a row in your db and connected UIs instantly update. Real-time UI isn’t a new concept, but sync engines like ElectricSQL turn Postgres into the source of truth for the sync, simplifying the mental model needed to build reactive updates - just write to the database!

These are systems-level changes that make a real difference, although we still have work to do.

Read the full engineering writeup here: neon.com/blog/recent-...

7/ Better sharding policy

We now shard earlier and more gradually: 4 shards at 16 GB, 16 at 2 TB. This improved ingest performance by 17.5% for mid-size tenants.

More info on sharding: neon.com/blog/how-we-...

6/ Parallel S3 uploads

Uploads no longer block layer writes. Upload throughput doubled, and ingest time in our benchmarks dropped.

5/ Smarter compaction

We now prioritize L0 compaction to avoid read amplification during ingestion. This dropped p99 read amplification by >50%.

4/ Scan prefetching

Instead of fetching one page at a time, Pageservers now prefetch up to 100. In our benchmarks, schema finalization dropped from 10h to 2.5h.

3/ Faster disk writes

Safekeepers now use sparse files and fewer flushes. WAL write throughput improved by 230%.

2/ WAL compression

WAL is now compressed with Zstd before it’s sent to Pageservers. This cut Safekeeper → Pageserver bandwidth by ~70%.

1/ Sharded ingest

We stopped broadcasting WAL to all shards and now route writes only where they’re needed. Less network traffic, lower CPU, and ~2× faster ingestion for large tenants.

To finish up, here’s a demo: fyi.neon.tech/pgsearch

Search over 50 million records using pg_search and Neon!