How useful is this structured log ingestion benchmark for choosing a platform?

November 18, 2026 5 answers
Asked by 7g!mavreni_284

A vendor-affiliated benchmark compares VictoriaLogs, Loki, Elasticsearch, OpenSearch, SigNoz and ClickStack using structured Kubernetes-style logs on identical 8-core machines. How much should the ingestion results influence a platform choice, and what else should I test?

Answers

Zavreno-5h

It's a useful starting point, not a buying decision. The benchmark ramps ingestion until each system can no longer keep up, which tells you something about throughput under that particular setup. The author disclosed working on VictoriaLogs, and VictoriaLogs was not the highest-ingesting system in that test.

I'd rerun the available benchmark with your own log shapes, field counts and traffic patterns. Check the configurations too. Identical machines don't necessarily mean equivalent indexing choices, durability settings or operational trade-offs.

5r+fablethimble_629:

And test the rate you need to sustain, not just the maximum. A peak number alone won't tell you what happens during a burst.

Reply to this answer
9h!flindaro_351

Ingestion is only half the story. Run your actual searches and alerts against a realistic amount of data, especially queries that filter on structured fields. A system that writes quickly may still need more query-time CPU or narrower time ranges to return results quickly.

For Loki, object storage can be a major draw, but field-heavy queries may involve trade-offs. Depending on the workload, deriving metrics from logs can make dashboards more practical than repeatedly querying the raw logs.

Reply to this answer
_f92~drivnelo.7

Don't overlook migration work. VictoriaLogs accepts the Loki push API, so changing where logs are sent may be fairly straightforward. Queries are another matter: VictoriaLogs uses LogsQL rather than LogQL, so dashboards and alerts may need rewriting. You don't necessarily have to switch everything at once.

7g!mavreni_284:

That's a useful distinction. I'll include the dashboard and alert rewrites in the migration estimate.

Reply to this answer
5t~quorlixa_829

I'd put operating cost and maintenance next to throughput. Storage choices, query resources and the effort to run and upgrade a system can matter more than the winner of a single ingestion test. The discussion includes favorable reports about VictoriaLogs, but those aren't a substitute for pricing and testing your own setup.

Reply to this answer
_m62~zunvexa.8

Keep metrics and logs as separate decisions if you can. This comparison is about structured log ingestion, so it doesn't answer how VictoriaMetrics, Mimir or Thanos will perform for your metrics workload. Benchmark that separately rather than assuming the log results carry over.

Reply to this answer

Add your answer

Answers are reviewed before publication.

Related Discussions