Don't forget de filesystem layer : Unless you use direct I/O or 
rawdevices, filesystem cache management will add extra latency that will 
be counted by Oracle as I/O service time : Oracle sends I/O via 
read/write/pread/pwrite/aoiread/aiowrite syscalls which are at last 
manager by the filesystem layer .
Also filesystem mappings can produce Oracle I/O breakups that could also 
add latency.

At last, all these kernel latencies which are not counted as I/O time by 
VxVM will be counted as I/O time by Oracle.


Craig Simpson a écrit :
>
> The Volumes below pretty much explain their purpose. 
>
> What I am running into is my DBA's say that they are not getting the
> read/write times that I see via vxstat. 
>
> What are "AVG TIME(ms) READ  WRITE" ???
>
> These mean the time it takes to a read or write to start, or is it the
> time a read or write takes to complete? 
>
> >From below it looks likes Oracle is getting really good read/write
> times. 
>
> Is it fair to say that, "Hey DBA'S this is what you are getting
> read/write Veritas says so" ???
>
>
>
>                       OPERATIONS          BLOCKS           AVG TIME(ms)
> TYP NAME              READ     WRITE      READ     WRITE   READ  WRITE
>
> Tue Apr 01 00:36:22 2008
> vol data01vol          735      1762     24400     41497    2.9    0.8
> vol data02vol          390      1374      6544     25307    7.0    0.7
> vol redo1              533     11635   1083396    233640    0.9    1.1
> vol redo2                4     11635         4    233640    2.5    1.1
>
> Tue Apr 01 00:38:22 2008
> vol data01vol          381     10665     18624    297160    4.1    3.1
> vol data02vol          636      1783     10432     31760    11.1    3.8
> vol redo1              495     16961   1011046    200950    0.2    1.1
> vol redo2                0     16960         0    200934    0.0    1.0
>
> Tue Apr 01 00:40:22 2008
> vol data01vol          333     13108     17504    313880    5.2    0.7
> vol data02vol          328      1745      5568     31208    15.6    0.8
> vol redo1                0      4773         0     63931    0.0    0.7
> vol redo2                0      4773         0     63931    0.0    0.7
>
> Tue Apr 01 00:42:22 2008
> vol data01vol          310      9324     17184    227240    3.3    0.7
> vol data02vol          334      1524      5632     27338    10.7    0.6
> vol redo1                0      4042         0     46028    0.0    0.8
> vol redo2                0      4042         0     46028    0.0    0.7
>
> --
>  
> Craig Simpson
> Visto Corporation
> Operations Team
> [EMAIL PROTECTED]
>
>
> "In the circle the beginning and the end are common"
>  ~ Heraclitis (540-480BC)
>
>  
>
> _______________________________________________
> Veritas-vx maillist  -  Veritas-vx@mailman.eng.auburn.edu
> http://mailman.eng.auburn.edu/mailman/listinfo/veritas-vx
>
>
>   


-- 
Sebastien DAUBIGNE
[EMAIL PROTECTED] - +33(0)5.57.89.31.09
AtosOrigin Infogerance - AIS/D1/SudOuest/Bordeaux/IS-Unix

_______________________________________________
Veritas-vx maillist  -  Veritas-vx@mailman.eng.auburn.edu
http://mailman.eng.auburn.edu/mailman/listinfo/veritas-vx

Reply via email to