Veritas-bu

Re: [Veritas-bu] Buffer settings LTO II vs LTO IV

2009-01-17 23:24:39
Subject: Re: [Veritas-bu] Buffer settings LTO II vs LTO IV
From: "bob944" <bob944 AT attglobal DOT net>
To: <veritas-bu AT mailman.eng.auburn DOT edu>
Date: Sat, 17 Jan 2009 23:10:18 -0500
> In an environment with HP LTO II's, I've been using for some 
> time and with good results:
>  
> NET_BUFFER_SZ (65536)
> SIZE_DATA_BUFFERS (262144)
> NUMBER_DATA_BUFFERS (32)
>  
> My question is has anyone seen the need, or any performance 
> improvement by changing these setting after implementing a 
> system with LTO IV's drives?
>  
> The drives in question are HP, but I'll look forward in 
> hearing about any experience that resulted in a performance 
> fix or increase.

This isn't quite the question you asked, but IMO, the _only_ way to know
optimal settings for a given environment is to test.  

a) find the best your media server(s) can do with realistic data under
ideal conditions (vary number, size buffers and mpx level to find best
setting for each)

b) test with repeatable data and conditions that replicate the
environment for which you are tuning, varying primarily network/comm
buffers and job load to try to reach levels found in a

That said, in my limited LTO4 experience (though not with HP's LTO4s)
32x256K is not enough to run LTO4s at native speed without relatively
incompressible data, a killer media server and a fair amount of
multiplexing.  In testing and real life, let the bptm logs guide.



_______________________________________________
Veritas-bu maillist  -  Veritas-bu AT mailman.eng.auburn DOT edu
http://mailman.eng.auburn.edu/mailman/listinfo/veritas-bu

<Prev in Thread] Current Thread [Next in Thread>