Catalogue
- Cards
- 165
- Records
- 486
- Workloads
- 19
- Corpus version
- 74847648e25a
17 of 486 records match.
| Kind | Record | Organisation | Year | Setup |
|---|---|---|---|---|
| Failure | Brief pod reduction during routine rollout saturated remaining sites, OOMed service-mesh sidecars, and triggered invalid-job runner retries | GitHub | 2026 | Multi-region |
| Decision | Scale ZippyDB keyspaces online via logical-to-physical shard indirection and coordinator-worker resharding | Meta | 2023 | Sharded |
| Decision | Offload Instagram-to-Threads follow-graph mirroring and notifications to rate-limited Async workers | Meta | 2023 | Sharded |
| Decision | Place channels on stateful Channel Servers by consistent hashing, spreading each team across all of them | Slack | 2023 | Multi-region |
| Decision | Deploy Gateway Servers in several regions so clients connect nearby, with draining to move users off a failed region | Slack | 2023 | Multi-region |
| Decision | Send transient events such as typing indicators through the same servers without saving them | Slack | 2023 | Multi-region |
| Decision | Migrate chat history from a single MongoDB replica set to Cassandra once data and index outgrew RAM | Discord | 2017 | Sharded |
| Decision | Bound Cassandra partition sizes with 10-day time buckets and Snowflake clustering keys | Discord | 2017 | Sharded |
| Decision | Delete corrupt edit/delete race rows when author_id is null and omit null columns on Cassandra writes | Discord | 2017 | Sharded |
| Failure | Millions of message tombstones in one channel triggered 10-second Cassandra stop-the-world GC pauses | Discord | 2017 | Sharded |
| Failure | Database contention filled Redis memory, and a full Redis could neither accept nor release jobs | Slack | 2017 | Sharded |
| Decision | Added Kafka in front of the Redis job queue instead of replacing Redis | Slack | 2017 | Sharded |
| Decision | Enqueue path acknowledges after the Kafka leader only, trading a small risk of lost jobs for latency | Slack | 2017 | Sharded |
| Decision | Split the monolithic Dispatcher into per-source Tailer adapters and a generic Scheduler service | Meta | Multi-region | |
| Decision | Switch workload transport from pull to RPC push from Tailers to the Scheduler | Meta | Multi-region | |
| Decision | Support both FOQS queues and Scribe streams as first-class storage backends for async jobs | Meta | Multi-region | |
| Decision | Offload failed stream items to a controlled-delay service to keep At-Least-Once retries from clogging checkpoints | Meta | Multi-region |