Catalogue
- Cards
- 165
- Records
- 486
- Workloads
- 19
- Corpus version
- 74847648e25a
33 of 486 records match; the first 20 are shown.
| Kind | Record | Organisation | Year | Setup |
|---|---|---|---|---|
| Decision | Offer Best-Effort counters on EVCache alongside Eventually Consistent global counters instead of probabilistic sketches | Netflix | 2024 | Multi-region |
| Decision | Log idempotent increments in Cassandra TimeSeries and roll up over a lagged immutable window | Netflix | 2024 | Multi-region |
| Decision | Trigger rollups without distributed locks via per-instance in-memory queues, Set coalescing, lastWriteTs and read self-remediation | Netflix | 2024 | Multi-region |
| Decision | Move I/O-bound telemetry receptors and processor from AWS Lambda and CloudFront to Cloudflare Workers | Cloudflare | 2024 | Multi-region |
| Decision | Replace self-hosted ClickHouse on EC2, Kinesis, and multi-service AWS alerting with Workers Analytics Engine and Durable Objects | Cloudflare | 2024 | Multi-region |
| Decision | Migrate incrementally via strangler-fig Queue mirroring and 30-day outside-in telemetry dual writes | Cloudflare | 2024 | Multi-region |
| Failure | Hidden dependencies on PDX-04-only Kafka and ClickHouse broke HA control-plane services during a full-facility power loss | Cloudflare | 2023 | Multi-region |
| Failure | Failing over control-plane services to European disaster recovery sites triggered an API thundering herd | Cloudflare | 2023 | Multi-region |
| Failure | Faulty breakers and a full cold bootstrap of PDX-04 prolonged recovery; log push data from the outage was not recovered | Cloudflare | 2023 | Multi-region |
| Decision | Run core control plane active-active across three Oregon facilities while excluding logging from the HA cluster | Cloudflare | 2023 | Multi-region |
| Failure | Unintended automatic systemd update deleted Cilium CNI routes across five independent regions at once | Datadog | 2023 | Multi-region |
| Decision | Separate live telemetry intake and serving systems from historical storage | Datadog | 2023 | Multi-region |
| Decision | Use statically sharded share-nothing stores for primary telemetry and quorum-based stores for metadata | Datadog | 2023 | Multi-region |
| Decision | Stop using DynamoDB and serve key/value lookups, ingest and rate limiting from one fixed-size SingleStore cluster | Fathom Analytics | 2021 | One database |
| Decision | Keep database connections open by moving the dashboard and queue off per-request Lambda connections | Fathom Analytics | 2021 | One database |
| Failure | A bug in per-tenant limit enforcement let one tenant overload a shared cluster, and error handling made the overload spread | Grafana Labs | 2021 | Multi-region |
| Decision | Share metrics clusters between tenants under per-tenant data and rate limits, and size clusters partly from those limits | Grafana Labs | 2021 | Multi-region |
| Failure | A viral customer's pageviews backlog the shared queue for all sites while the database runs out of disk IOPS | Fathom Analytics | 2020 | One database |
| Failure | Unindexed backlog tables and a text-typed site ID kept the database slow after the disk was fixed | Fathom Analytics | 2020 | One database |
| Decision | Cap concurrent queue workers on purpose to protect the database, accepting a bigger backlog | Fathom Analytics | 2020 | One database |
13 more records. Sign in to see all 33.