Methodology
Testing Approach
- We automated the benchmarking scripts to create reproducible benchmark results at any time.
- The client workload generator running InfluxDB Telegraf agents produces the data ingestion workload for databases under comparison. We chose Telegraf as the data collection agent of choice for familiarity, ease of use, and ability to scale. Telegraf is representative of the majority of our customer workloads.
- To detect maximum ingest capacity, the test increases the number of clients until a degradation is observed in the ingestor performance as reflected by CPU and memory usage on the ingestors.
- The test executes queries sequentially for various time ranges and schedules during ingest to simulate actual workload conditions. Future benchmark testing plans include adding parallel execution of query workloads.
Tested Configurations
InfluxDB 3.0 (The tests were performed against Cloud Dedicated.)
- We deployed the InfluxDB Cloud Dedicated (v3.0) XS cluster using AWS EKS (Elastic Kubernetes Service) to run as a 3-node (1 ingestor, 1 querier, and 1 compactor) cluster in AWS.
- Cost to end user: $60K/yr
InfluxDB OSS 1.8 (The tests were performed against InfluxDB OSS. The tests ran in the same class as Cloud Dedicated, on the biggest available instance.)
- We deployed the InfluxDB Open Source v1.8 on m5.24xlarge instance in AWS. Storage is backed by 10 TB General Purpose SSD (gp3).
- Total CPU: 96vCPU
- Total Memory: 384GiB RAM
- Additional AWS Services used: None
- AWS Infrastructure cost to end user: ~$60K/yr
Tested Dataset
We designed the data for this benchmark to simulate an application performance monitoring workload using Telegraf input plugins to collect a variety of metrics and measurements. The data schema produced by the Telegraf plugins used in this test is found
here.
- Total values in dataset: 372 billion
- Dataset duration: 24 hours
- Measurement interval: 10 seconds
Tested Queries
The benchmark automated test framework used the following set of queries to observe query latencies for the products under observation:
