On Tue, 2008-04-29 at 22:29 +0800, Adrian Chadd wrote:
> On Wed, Apr 30, 2008, Amos Jeffries wrote:
>
> > >The metrics depend on the workload, but in general you need at least
> > >rate/throughput, response time, hit ratio (if caching), and errors. No
> > >single metric is meaningful. You can optimize one and "freeze" others at
> > >reasonable levels and workload often do that.
>
> > Thanks Alex. That was pretty much spot on what I was hoping to hear back.
> > From what we've been hearing the two key rates will be req/sec (at any
> > low response time) and hit ratio.
>
> Don't forget other key things which affect scalability, such as concurrent
> connection count and memory usage.
The number of active concurrent connections is just request rate
multiplied by response time. The number of idle persistent connections
can be a metric.
I agree that reporting memory/CPU/disk/NIC usage would be useful.
Alex.
Received on Tue Apr 29 2008 - 14:51:12 MDT
This archive was generated by hypermail 2.2.0 : Wed Apr 30 2008 - 12:00:07 MDT