You are viewing a plain text version of this content. The canonical link for it is here.
Posted to user@cassandra.apache.org by Brian Burruss <bb...@real.com> on 2009/12/16 20:03:27 UTC
read/write latency definition
what exactly is the definition of latency as reported by tpstats? is it the complete time spent in the cassandra server, minus sending response to client? that would be nice if it is ;)
if it isn't, what metrics can i use to account for the time in the server? trying to account for time so i can determine best course for optimizing.
thx
Re: read/write latency definition
Posted by Jonathan Ellis <jb...@gmail.com>.
On Wed, Dec 16, 2009 at 1:03 PM, Brian Burruss <bb...@real.com> wrote:
> what exactly is the definition of latency as reported by tpstats?
You mean cfstats, right?
That's the latency from when the request arrived at the right server,
to read the requested data.
> is it the complete time spent in the cassandra server, minus sending response to client?
Nope. We track that globally in StorageProxy, and it is exposed via
JMX, but not via nodeprobe. (Adding that would be easy tho, if you
wanted to do it.)
-Jonathan