In-memory dimensional time series database.
3.6k
Stars
354
Forks
12
Open issues
26
Contributors
AI Analysis
Atlas is Netflix's in-memory dimensional time series database backend for managing and querying large-scale metric data with dimensional tags. It is purpose-built for high-cardinality observability and monitoring workloads at scale, serving Netflix's internal infrastructure and organizations requiring specialized time series storage with dimensional indexing. This tool is not a general-purpose relational database and is best suited for teams building monitoring and telemetry systems who under...
Inferred from signals mentioned in the README (tests, CI, type safety) — not a review of the actual code.
AI's overall editorial judgment — not an average of the bars above, can weigh other factors too.
Netflix's in-memory time series database for dimensional metrics at scale
Atlas is an in-memory dimensional time series database designed for high-cardinality metrics workloads, originally built to solve Netflix's internal observability at scale. It stores time series data with multiple dimensions (tags) in memory, enabling fast aggregation and graphing. Primary users appear to be organizations with similar infrastructure-scale monitoring needs. Adoption remains largely confined to companies explicitly running Netflix-style observability stacks.
Atlas was created by Netflix in 2014 to address limitations in existing time series solutions for handling high-cardinality dimensional metrics. It evolved from Netflix's internal operational requirements and remains maintained as open source, though primarily documented and adopted within Netflix's ecosystem and similar large-scale operations.
The project gained 3,554 stars over 12 years, suggesting steady but modest adoption. No stars gained in the last 7 days (as of 2026-07-08). Growth appears driven by organizations scaling observability infrastructure rather than broad market adoption. The project maintains a long, stable history without explosive growth phases, consistent with a specialized infrastructure tool serving a defined niche.
Netflix's own use is documented (it is their internal monitoring backend). Mailing list exists, suggesting user community engagement. However, specific evidence of adoption outside Netflix or case studies from other organizations is not provided in README. Adoption not verified beyond Netflix's internal deployment.
Based on README, Atlas is an in-memory dimensional time series backend. Likely written in Scala for JVM deployment. Appears designed for time series ingestion, storage, and dimensional aggregation queries. README does not detail sharding, replication, or fault-tolerance strategies; architecture specifics not fully documented in available excerpts.
not documented in README
Last push 2026-07-03 (5 days before evaluation date) indicates active maintenance. Repository created 2014-08-05 shows 12 years of continuous history. Maintained by Netflix, not a solo project. Regular maintenance pattern suggests stability but not evidence of accelerating development velocity.
ADOPT IF: your organization runs Netflix-scale infrastructure with high-cardinality dimensional metrics, has Java/Scala expertise, and can operate an in-memory time series backend with custom operational overhead. AVOID IF: you need broad ecosystem support, want managed deployment, or require time series solutions that work well in resource-constrained environments or cloud-native defaults. MONITOR IF: you are evaluating observability backends for large infrastructure and can tolerate the learning curve of a specialized tool with smaller community than Prometheus or InfluxDB.
Independent dimensions
Mainstream potential
3/10
Technical importance
7/10
Adoption evidence
4/10
- Adoption limited to specialized use case (high-cardinality metrics at scale); may not be suitable for typical SaaS or small-to-medium infrastructure monitoring.
- In-memory storage model requires significant RAM and may not align with modern cost-optimized infrastructure patterns.
- Community size and third-party integrations appear smaller than mainstream alternatives; dependency on Netflix for long-term maintenance roadmap.
- Documentation and examples appear focused on Netflix use patterns; onboarding friction likely higher for teams outside that context.
- Lack of managed/cloud-hosted offering may disadvantage adoption compared to SaaS alternatives.
Atlas will likely remain a specialized backend for organizations with Netflix-comparable infrastructure complexity and high-cardinality metric needs. Adoption will continue at modest, stable levels. Unlikely to achieve mainstream market dominance, but will persist as a reference implementation and tool for its niche.
Explore similar
Newsletter
Get analyses like this every Monday
Free weekly digest of the most interesting open-source discoveries.
Languages
Information
- Language
- Scala
- License
- Apache-2.0
- Last updated
- 1w ago
- Created
- 145mo ago
- Analyzed with
- anthropic/claude-haiku-4-5
Stars over time
Contributors over time
Top 100 contributors only — repos with more will plateau at 100.
Open issues
How to use atlas-postgres module as the database backend?
Support constants that could be used along with a grouping
automatically publish release assets
custom palette that allowed mapping legend values to colors
Support time zones for math time functions
Top contributors
Similar repos
| Repository | Stars | Week Δ | Language | Score | Updated |
|---|---|---|---|---|---|
|
Netflix/atlas
this
|
3.6k | — | Scala | 8/10 | 1w ago |
 apache/atlas
|
2.1k | — | Java | 7/10 | 19h ago |
 kairosdb/kairosdb
|
1.8k | — | Java | 7/10 | 4mo ago |
|
apache/iotdb
|
6.4k | — | Java | 8/10 | 7h ago |
6,359 stars, Java-based, time series optimized. Appears to target broader IoT and time series use cases; not specifically dimensional metrics. Higher star count may reflect broader appeal but does not indicate feature parity or suitability for Netflix-scale observability.
1,759 stars, Java, time series database. Similar era and architectural inspiration (HBase-backed in some modes). Lower adoption suggests either narrower use case or loss of relevance over time.
2,116 stars, Java, metadata/lineage management. Different problem domain (data governance vs. metrics time series); not a direct competitor despite similar name recognition.
Not listed in similar repos but implicit competitor in observability space. Prometheus emphasizes pull-based collection and multi-dimensional labels; more widely adopted in cloud-native environments. Atlas emphasizes in-memory performance for high-cardinality metrics.
Not listed but commonly compared in time series space. InfluxDB emphasizes ease of use and broader ecosystem; Atlas emphasizes dimensional query performance at Netflix scale.