Nikita | Scaling Postgres
@nikitabase
Followers
11K
Following
7K
Media
243
Statuses
5K
CEO @neondatabase. Best Postgres for developers and vibe coders.
San Francisco, CA
Joined June 2010
Database architecture thread. Technical. There has been several startups building an operational relational databases focused on OLTP with a shared nothing architecture. @neondatabase is using a different approach - shared storage. What's the difference?
15
73
504
@neondatabase thanks for building an awesome database product! Excited to build more things on neon soon π
Agent builders, be like @dyad_sh - donβt stop your time travel at the code layer. Version both code and data by pairing Git commits with Neon branches:
0
1
7
Twitter's former CEO just raised $100 million for his new AI startup
parallel.ai
Parallel announces $100M Series A at $740M valuation to build web infrastructure for AI agents. We're creating better search for the web's second user.
6
5
62
Neon's snapshots are now live for everyone! Schedule automatic daily, weekly, or monthly database backups. Capture your data's exact state effortlessly. Big news from @neondatabase! source: @neondatabase
0
1
1
Branching saves money. It's faster, simpler, and more secure
Just shipped: you can now select Anonymized data when creating a Neon branch Youβll get a perfect copy of your database (schema and data) BUT with your PII columns already masked according to the rules you choose:
2
1
15
now you can branch with security, (PII not included)
Just shipped: you can now select Anonymized data when creating a Neon branch Youβll get a perfect copy of your database (schema and data) BUT with your PII columns already masked according to the rules you choose:
0
1
4
Just shipped: you can now select Anonymized data when creating a Neon branch Youβll get a perfect copy of your database (schema and data) BUT with your PII columns already masked according to the rules you choose:
3
2
34
Our customers always loved branching, but have always been concerned about PII. This is no more. Anonymized branches are here and you have tight control over PII.
Just shipped: you can now select Anonymized data when creating a Neon branch Youβll get a perfect copy of your database (schema and data) BUT with your PII columns already masked according to the rules you choose:
1
1
25
Edge latency between database and serverless compute is becoming a non issue
Nature is healing: β’ Neon ππ (TCP): 3.00ππ avg π β’ Neon HTTP: 4.50ππ avg β’ Neon WebSocket: 3.90ππ avg π Fluid solves the connection pooling issues of serverless. And works *best* with battle-tested clients like ππ. Itβs the ideal backend runtime. No lock-in,
2
1
22
1
1
8
Nature is healing: β’ Neon ππ (TCP): 3.00ππ avg π β’ Neon HTTP: 4.50ππ avg β’ Neon WebSocket: 3.90ππ avg π Fluid solves the connection pooling issues of serverless. And works *best* with battle-tested clients like ππ. Itβs the ideal backend runtime. No lock-in,
Fluid compute changes how serverless works on Vercel: - Classic serverless: Connection to db must be established on every run, so http/websocket were faster - Fluid: Can establish a db connection once and keep it open across many function calls, so pg tcp is now fastest
15
11
212
Today's Future Data Systems Seminar Speaker: Cheng Chen will present how @mooncakelabs extends PostgreSQL to support @ApacheIceberg. Zoom talk open to public at 4:30pm ET. YouTube video available after:
db.cs.cmu.edu
Apache Iceberg is great for large-scale analytics, but it was built for... Read More +
0
6
60
@nicoritschel @nikitabase @neondatabase Yes to remote. I used to use local PG for dev, but I've found the Neon UI and branching faster, so I just use that now. Of course I can undo commits, but in dev, I treat it like git now: I'm not usually reverting commits... I just reset a branch. Same with the DB.
0
1
2
The Neon MCP server is open source and you can check it out on GitHub! - stdio transport (released a year ago) - streaming http transport w/ OAuth flow - read-only mode (PR open)
3
1
15
This is a good question. We need escalating benchmark complexity. Start with micro benchmarks that measure latency, throughput and jitter for reads and writes. Make sure they measure cliffs: working sets exceeding cache sizes and triggering background processes (checkpoints,
The database community needs better benchmarks. TPCC is widely used for OLTP, but was published in 1992. Who's working on this? What do you want to see in a modern OLTP benchmark?
0
0
15
@nikitabase @nicoritschel @neondatabase Reset from parent really is a killer feature. I use it constantly. An incredibly practical Undo button when trying schemas ideas and ending up with junk data.
1
1
2
Hmm we have a warm pool for creating instances and branches. So must be a pool miss. Please file a support ticket with details and we will improve!
1
0
0
Most modern database providers embraced branching, but charge ~$10/month/branch. @neondatabase makes branches nearly free since dev workloads that don't consume much. They are instant to create and have a full copy of your data.
5
0
49
@sebastiankehle_ i like neon bc it's serverless postgres and the branches are great, not sponsored btw
0
1
4
"Initializing" for is the new anti pattern of modern dev infrastructure. Developers expect "instant" for all provisioning operations.
0
0
5
Serverless!
Friday Update: You can now have up to tΜΆwΜΆeΜΆnΜΆtΜΆyΜΆ ππ΅πΆπΏππ projects on the Neon Free Plan
1
1
31