How can I bring down Log Analytics costs for AKS?

November 18, 2026 4 answers
Asked by Nerquavi-4m

Our AKS clusters send telemetry to a Log Analytics workspace that now costs more than the node pools. ContainerLogV2 is the biggest table, followed by Perf and KubeEvents. Where would you start cutting costs without losing useful diagnostics?

Answers

2g!cobblewhistle_759

Start with ContainerLogV2 since it is driving most of the usage. Check which workloads and namespaces are producing the volume, then reduce overly chatty application logs or exclude data you genuinely don't need through the Container Insights ConfigMap or data collection rule.

Error-only logging can cut a lot, but it may leave you short of context during an incident. Consider which services need warnings or other diagnostic information before making that the rule everywhere.

_c46~ferrulark.8:

Perf and KubeEvents are worth reviewing too, just after the biggest source. Metrics and event volume can add up even when workloads aren't doing much.

Reply to this answer
_r83~plivaro.4

Look at whether ContainerLogV2 fits the Basic logs plan. It can have a lower ingestion price, but comes with more limited query capabilities and shorter interactive retention. Check the current pricing and table eligibility, then test the queries your team relies on before changing the plan.

Also review retention separately. Keeping less data may reduce storage costs, but shortening retention won't fix a large ingestion bill by itself.

Nerquavi-4m:

That distinction helps. I'll check how often we query older container logs before changing retention or the table plan.

Reply to this answer
p.594!brindlewidget

Have a look at the OpenTelemetry collection options too. They can give you more control over what gets collected and sent to Log Analytics, which is useful if the current Container Insights settings are broader than you need.

Reply to this answer
5t!mivnori_263

If filtering and table plans still leave the bill too high, compare another logging setup, such as an ELK stack or a Grafana-based option. Don't just compare ingestion prices though. Factor in storage, operations, and the effort of maintaining the replacement.

Reply to this answer

Add your answer

Answers are reviewed before publication.

Related Discussions