Hacker Newsnew | past | comments | ask | show | jobs | submit | jonahberquist's commentslogin

Discord | Senior Software Engineer - Database Infrastructure | SF Bay Area (Hybrid)

Tech stack: Rust | ScyllaDB | k8s | Python | Postgres | Elasticsearch | Terraform

Role: Build and operate distributed database systems and data platform services that store all of the data for Discord's 200M+ users.

We're solving interesting problems around performance, reliability, and scalability at one of the largest communication platforms. Looking for strong distributed systems experience, familiarity with databases like Cassandra/ScyllaDB/Postgres, and comfort working in Rust or similar systems languages.

https://discord.com/jobs/8200328002


This is more dependent on parallelism of queries rather than parallelism of connections. Having a ton of relatively idle connections won't increase the query latency.

For the related topic, in https://planetscale.com/blog/one-million-queries-per-second-... we touched on saturating shards through increased parallelism resulting in latency, and then also being an indication of needing to scale more horizontally.


We used https://github.com/Percona-Lab/sysbench-tpcc, which the folks at Percona did a good write up of here https://www.percona.com/blog/2018/03/05/tpcc-like-workload-s.... For these tests, we ran with --tables=1 and --scale=1000.


Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: