VictoriaMetrics and Grafana Mimir are time series databases with support of mostly the same protocols and APIs. However, they have different architectures and components, which makes the comparison more complicated. In the talk, we'll go through the details of the benchmark where I compared both solutions. We'll see how VictoriaMetrics and Mimir are dealing with identical workloads and how efficient they’re with using the allocated resources.
The talk will cover design and architectural details, weak and strong points, trade-offs, and maintenance complexity of both solutions.
Grafana Mimir and VictoriaMetrics_ Performance Tests.pptx
1. Grafana Mimir and VictoriaMetrics
Performance Tests
Roman Khavronenko | github.com/hagen1778
2. Roman Khavronenko
Co-founder of VictoriaMetrics
Software engineer with experience in distributed systems,
monitoring and high-performance services.
https://github.com/hagen1778
https://twitter.com/hagen1778
3. The High Performance
Open Source Time Series Database & Monitoring Solution
120M+ downloads
11k stars in total
200+ contributors
Grammarly, CERN,
Open Cosmos, Wix
Semrush, Roblox
4.
5. Stats:
● Active time series: 1Bil
● Ingestion: 50Mil/s
Resources:
● CPUs: 7000
● RAM: 30TiB
● Replicas: 1500
23. Benchmark
Stats:
● Duration: 24h
● Ingestion: 360K samples/s
● Unique time series: 5.5Mil
● Churn rate: 1% each 10 min
● Total number of series over 24h: 13.6Mil
● Total number of samples over 24h: 31Bil
29. Disk space used
Mimir:
Replication: x3
Local FS: 300 GiB
Object storage: 150 GiB
-blocks-storage.tsdb.retention-period
To reduce disk size at local FS
compactor runs compaction jobs at:
2h, 12h and 24h
compactor deduplicates replicated data
VictoriaMetrics:
Replication: x2
Local FS: 50 GiB
Downsampling
-downsampling.period=30d:5m,180d:1h
Retention filters
-retentionFilter='{env="dev"}:3d'
-retentionFilter='{env="staging"}:30d'
-retentionPeriod=1y
No object storage support
Stores all replicated data
38. Benchmark - x5 increase in load
Stats:
● Duration: 3h
● Ingestion: 1.8Mil samples/s (x5)
● Unique time series: 29 Mil (x5)
● Churn rate: 1% each 10 min
● Total number of series over 3h: 32 Mil
● Total number of samples over 3h: 19 Bil