I was wondering what is the average OPS per instance per clustered server number.
The reason is to estimate the required number of servers to perform our required speed
This is dependent on your workload as well as your network speed. For example if you have all of your active items in memory you can easily get 200k+ ops/sec. CHeck out this blog post from Cisco.
If you can provide details on what your workload is like, your expected data size, and your maximum working data set size I can try to help estimate a good cluster size
© 2013 COUCHBASE All rights reserved.