On Thu, May 29, 2008 at 02:04:21PM -0700, eric kustarz wrote: > > > >Currently both tools gives different view of io. While filebench > >simulates real workload, it cannot show what vdbench shows. > >E.g. we were doing some hardware array tests and it turned out (using > >vdbench) that the array works in a really strange way: sometimes its > >IOPS jumps very high, sometimes it lows very much. There were > >a few other tests where vdbench shows how IOPS behaves in each > >second of workload. I am affraid filebench _currently_ cannot give us > >the same data. > > Good point. > > We actually have this via Xanadu, though i just tried it and it looks > like Xanadu is not working. > > Also, at the UCSC benchmarking conference last monday we kicked around > the idea of showing a distribution of results instead of just averages.
Averages can be really misleading. They don't show in depth data. I remember another hardware arrays (yes, two arrays) test where vdbench shows that one array is a bit faster (more IOPS) then the other. While filebench (which tries to simulate the real workloads) shows that that the other aray is slightly faster. What is true ? I don't know. But I wouldn't like to use just one tool. Both has its advantages and disadvantages. And _both_ give much wider picture of how the storage behaves in particular workload. And I am sure that if you include vdbench in /usr/benchmarks, there will be many who would like to use _both_, not just filebench. After all, one way of improving Solaris is giving a choice :-) And to comment the following sentence: "vdbench has no future" - let _me_ decide what is better for me. Regards przemol -- http://przemol.blogspot.com/ ---------------------------------------------------------------------- Tanie rozmowy! Sprawdz >>> http://link.interia.pl/f1e13 _______________________________________________ perf-discuss mailing list perf-discuss@opensolaris.org
