Why is Couchbase such a Hog?
Tue, 02/12/2013 - 08:05
I've been evaluating Couchbase 2.0 Enterprise in a VM environment with 5 nodes @ 2.0ghz 8 Core and 32GB of Ram. Even when the entire cluster is idle, doing nothing at all it looks like all of the servers are consuming quite a lot of CPU and RAM.
Top looks like memcached and beam.smp are the two primary hogs. Anyone else have these observations as well? Is there anything that can be done to make CB more efficient? Maybe I just don't know what's happening behind the scenes, it just seems a bit excessive compared to my Datastax Cluster...also doing nothing.
Hello,
First of all Couchbase Server is using as you know memcache to cache all the keys and some documents. This is how we can deliver content in a very fast way. So depending of the configuration of your cluster and its number of buckets you will have some RAM allocated to the cache.
Also Couchbase is doing things in the background especially to deal with the information on disk, for each bucket we have processes to check indexing, compaction of files... this is why you still have some activity.
That said I am curious to understand what is happening on your cluster, can you sare wit us:
- how many buckets?
- how many documents do you have? replicas?
- do you use views? replicated index?
- how much memory for each buckets?
If you want to understand what is going on your cluster and how to size it properly I am inviting you to read this blog post:
- http://blog.couchbase.com/how-many-nodes-part-1-introduction-sizing-couc...
Regards
Tug
@tgrall