Kiu/VictoriaMetrics
Kiu/VictoriaMetrics
1
0
Fork 0
mirror of https://github.com/VictoriaMetrics/VictoriaMetrics.git synced 2024-12-13 13:11:37 +01:00
Code Issues Actions 1 Packages Projects Releases Wiki Activity

docs/Cluster-VictoriaMetrics.md: remove incorrect information about the increase of workload on the remaining vmstorage nodes when one vmstorage node is unavailable

Browse Source 
If one out of 5 vmstorage nodes is unavailable, then the remaining 4 vmstorage nodes will recieve 1/5=20% increase of workload, not 5%.
If one out of 10 vmstorage nodes is unavailable, then the remaining 9 vmstorage nodes will receive 1/10=10% increase of workload, not 1%.

This is a follow-up for 458338afa5

Updates https://github.com/VictoriaMetrics/VictoriaMetrics/pull/6099
This commit is contained in:
Aliaksandr Valialkin 2024-04-17 10:06:39 +02:00
parent 3f618c0485
commit c6f0a37125
No known key found for this signature in database
GPG Key ID: 52C003EE2BCDB9EB
1 changed files with 17 additions and 3 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

20
docs/Cluster-VictoriaMetrics.md
View File

@ -458,10 +458,24 @@ Cluster performance and capacity can be scaled up in two ways:
General recommendations for cluster scalability:
- Adding more CPU and RAM to existing `vmselect` nodes improves the performance for heavy queries, which process big number of time series with big number of raw samples. See [this article on how to detect and optimize heavy queries](https://valyala.medium.com/how-to-optimize-promql-and-metricsql-queries-85a1b75bf986).
- Adding more `vmstorage` nodes (aka horizontal scaling) increases the number of [active time series](https://docs.victoriametrics.com/FAQ.html#what-is-an-active-time-series) the cluster can handle. This also increases query performance over time series with [high churn rate](https://docs.victoriametrics.com/FAQ.html#what-is-high-churn-rate). The cluster stability is also improved with the number of `vmstorage` nodes, since active `vmstorage` nodes need to handle lower additional workload when some of `vmstorage` nodes become unavailable. For example, during a node outage the rest of the nodes will pick up the load designated to unavailable node. So in case one node out of 5 is unavailable 20% of the load will be re-distributed across 4 remaining nodes which means each node will take 5% of the load. With 10 nodes in case of outage 10% of the load will be distributed across 9 remaining node, so around ~1% of load will be distributed across other nodes.
- Adding more CPU and RAM to existing `vmstorage` nodes (aka vertical scaling) increases the number of [active time series](https://docs.victoriametrics.com/FAQ.html#what-is-an-active-time-series) the cluster can handle. It is preferred to add more `vmstorage` nodes over adding more CPU and RAM to existing `vmstorage` nodes, since higher number of `vmstorage` nodes increases cluster stability and improves query performance over time series with [high churn rate](https://docs.victoriametrics.com/FAQ.html#what-is-high-churn-rate).
- Adding more CPU and RAM to existing `vmselect` nodes improves the performance for heavy queries, which process big number of time series with big number of raw samples.
See [this article on how to detect and optimize heavy queries](https://valyala.medium.com/how-to-optimize-promql-and-metricsql-queries-85a1b75bf986).
- Adding more `vmstorage` nodes (aka horizontal scaling) increases the number of [active time series](https://docs.victoriametrics.com/FAQ.html#what-is-an-active-time-series)
the cluster can handle. This also increases query performance over time series with [high churn rate](https://docs.victoriametrics.com/FAQ.html#what-is-high-churn-rate),
since every `vmstorage` node contains lower number of time series when the number of `vmstorage` nodes increases.
The cluster stability is also improved with the number of `vmstorage` nodes, since active `vmstorage` nodes need to handle lower additional workload
when some of `vmstorage` nodes become unavailable. For example, if one node out of 5 is unavailable, 20% of the load is re-distributed across 4 remaining nodes.
If one node out of 10 nodes is unavailable, 10% of the load is re-distributed across 9 remaining node.
- Adding more CPU and RAM to existing `vmstorage` nodes (aka vertical scaling) increases the number
of [active time series](https://docs.victoriametrics.com/FAQ.html#what-is-an-active-time-series) the cluster can handle.
It is preferred to add more `vmstorage` nodes over adding more CPU and RAM to existing `vmstorage` nodes, since higher number of `vmstorage` nodes
increases cluster stability and improves query performance over time series with [high churn rate](https://docs.victoriametrics.com/FAQ.html#what-is-high-churn-rate).
- Adding more `vminsert` nodes increases the maximum possible data ingestion speed, since the ingested data may be split among bigger number of `vminsert` nodes.
- Adding more `vmselect` nodes increases the maximum possible queries rate, since the incoming concurrent requests may be split among bigger number of `vmselect` nodes.
Steps to add `vmstorage` node: