Hacker News Reader

Comment by pbowyer | Hacker News Reader

They both have replication, which allows them to be multi-node.

With the bounds capped by a single writer. Unless you can shard the data and create a distributed database with manual sharding.

But yes. Postgres remains an amazing choice, especially with modern hardware, until you also have the money available to tackle said write throughput issue.

I think the point is that sharding won't really help that much since transactions will happen across all or most shards, and then you have certain accounts that will be more active than others.

if your sharding schema is designed properly you will avoid cross shard transactions.

How so? Any account may transact with any other account - regardless which shard it resides in.

TigerBeetle doesn't shard either.

Good catch.

Yes, sharding would kill performance under contention, which characterizes many OLTP workloads (e.g. top ten bestseller list on Black Friday, super stocks like NVIDIA, the big 4 banks on a switch, PhonePe/Google Pay on UPI etc)

Sharding is if anything a way to reduce the contention.

Transactions typically need to hit the same set of house accounts. For example, fee accounts, omnibus accounts etc.

Or a central bank switch may only have 4 major banks.

Or a system like UPI may have 85% of transactions flowing through 2 hubs. Say 50% through PhonePe. And say 35% through Google Pay.

Reply on news.ycombinator.com