high load average
we are new to membase, at the moment the cluster is working just fine, but we have pretty high load average...
2 webserver behind a load balancer
the membase is installed directly on the webservers, version 18.104.22.168
each webserver has 16 GB ram and a cpu with 2x AMD Opteron(tm) Processor 250
the cluster is defined to use 12 GB ram of each node
there is a membase bucket: RAM Quota: 16GB, Replication: disabled
RAM Used: 3.5%, Item Count: 283602, Memory Bytes Used: 603M
so the items are about 2K each in average
at the moment the web load balancer gets 10 requests per second in average, every web request makes a "get" and a "set" to the mambase cluster, so there are 20 operations per second on the cluster in average
actually not very much load... we expect up to a couple of thousend request per second in the future...
what we are very worried about is the high load average at the moment on the nodes, without being requested frequently...
the load average is pending between 1.5 and 3.5, at the same time the CPU is used just at about 10%
what are we doing wrong? any suggestions?