[ADSM-L] IO profile of TSM storage
2012-06-29 03:00:58
Hi
I'm in the process of revisiting the TSM storage IO requirements and
thought I'd just get a heads up for any previous experience, corrections
etc.
From what I can gather in the documentation
* File based storage pools are read/written in 256KB chunks
* DB is read and written in 64KB chunks.
But....
what is the profile of active logs? archive logs?
Is there a way to increase the IO depth/concurrency when TSM
reads/writes to the storage pool etc
Does MOVESIZETHRESH and TXNBYTELIMIT offer much improvement?
Is there any way to determine if the DB/DB logs are slowing the
streaming from the Disk pools?
Do you know what the equiv "fio" would be to generate the same load
pattern as TSM. Using fio I was able to push a LUN to 300MB/s sequential
read but with TSM I'm getting about 60-70MB/s for a migration from file
to an LTO5 drive. Disk is not contended CPU is fine....any tips?
Grant
|
<Prev in Thread] |
Current Thread |
[Next in Thread> |
- [ADSM-L] IO profile of TSM storage,
Grant Street <=
|
|
|