Hi,
I currently have at my disposal a NetApp VTL 600 for testing purposes.
Setup :
It's a single shelf VTL containing 14 500GB SATA disks including 1
hotspare.
I have a Windows server attached via FC, TSM 5.4 installed on the
server.
The VTL is configured to be a NetApp library and it's simulating to have
4 IBM LTO3 drives.
Installation :
The default TSM device drivers that come with TSM contain the driver for
this 'netapp' library and of course the LTO3 drives pose no problem
since I can use the default IBM drivers for those as well.
My findings so far :
I inserted 50 virtual tapes in the VTL and started a label libv command
with a checkin=scratch.
This took a whole 30 somewhat secondes to complete..50 "tapes" labeled
and checked in.
It looks like TSM just can't interface with a library (be it virtual of
not) faster than this.
An audit libr without a checklabel=barcode seems to mount the first
virtual tape and then just sit there doing nothing.
DB backup (ok, it's a small db, but still)
>From the actual start of the backup to the point where the volume is
opened takes 1 second.
The whole thing for giving the command to the completion and dismount is
done in 33 secondes.
15-2-2008 13:21:16 ANR2017I Administrator ADMIN issued command: BACKUP
DB type=full devclass=LTOCLASS1
15-2-2008 13:21:24 ANR0984I Process 236 for DATABASE BACKUP started in
the BACKGROUND at 13:21:25.
15-2-2008 13:21:24 ANR2280I Full database backup started as process 236.
15-2-2008 13:21:24 ANR0405I Session 845 ended for administrator ADMIN
(WinNT).
15-2-2008 13:21:25 ANR8337I LTO volume 00001L7 mounted in drive
MT1.1.0.2 (mt1.1.0.2).
15-2-2008 13:21:25 ANR0513I Process 236 opened output volume 00001L7.
15-2-2008 13:21:25 ANR1360I Output volume 00001L7 opened (sequence
number 1).
15-2-2008 13:21:33 ANR4554I Backed up 7040 of 14162 database pages.
15-2-2008 13:21:44 ANR4554I Backed up 14080 of 14162 database pages.
15-2-2008 13:21:48 ANR1361I Output volume 00001L7 closed.
15-2-2008 13:21:48 ANR0515I Process 236 closed volume 00001L7.
15-2-2008 13:21:49 ANR4550I Full database backup (process 236) complete,
14162 pages copied.
15-2-2008 13:21:49 ANR0985I Process 236 for DATABASE BACKUP running in
the BACKGROUND completed with completion state SUCCESS at 13:21:50.
15-2-2008 13:21:49 ANR8336I Verifying label of LTO volume 00001L7 in
drive MT1.1.0.2 (mt1.1.0.2).
15-2-2008 13:21:49 ANR8468I LTO volume 00001L7 dismounted from drive
MT1.1.0.2 (mt1.1.0.2) in library LB1.0.0.2.
13:25:18 localhost : q stgpool
Storage Device Estimated Pct Pct High Low Next
Stora-
Pool Name Class Name Capacity Util Migr Mig Mig ge Pool
Pct Pct
----------- ---------- ---------- ----- ----- ---- ---
-----------
ARCHIVEPOOL DISK 0.0 M 0.0 0.0 90 70 DISKPOOL
BACKUPPOOL DISK 0.0 M 0.0 0.0 90 70 DISKPOOL
DISKPOOL DISK 20 G 31.3 31.3 100 99 LTOPOOL1
LTOPOOL1 LTOCLASS1 105,808 G 0.1 0.8 90 70
SPACEMGPOOL DISK 0.0 M 0.0 0.0 90 70
13:32:11 localhost : show transferstats diskpool
Statistics for last migration from pool DISKPOOL
Start date/time: 02/15/2008 13:27:36
Elapsed time: 165 seconds
Total wait time: 1 seconds
Number of participating processes: 4
Total duration of all processes: 165 seconds
Total physical files: 166
Total logical files: 22003
Total bytes: 6572945408
Average logical files per physical file: 132.5
Average physical file size: 38668.0 KB
Number of batch/file transactions ended: 15
Number of batch transactions aborted: 0
Number of file transactions started: 0
Number of file transactions aborted: 0
No storage pool backup information available for DISKPOOL.
Guess we got 38MB/s on that migration from a slow internal disk diskpool
-> VTL.
A funny this is that you can set the size of the virtual volumes in the
VTL independent of the drive type you select and configure in TSM.
I selected a 10GB data limit per volume on these LTO3 virtual volumes
and this is what I get :
13:37:00 localhost : q vol
Volume Name Storage Device Estimated Pct
Volume
Pool Name Class Name Capacity Util
Status
------------------------ ----------- ---------- --------- -----
--------
00001L3 LTOPOOL1 LTOCLASS1 9,510.2 100.0
Full
00001L4 LTOPOOL1 LTOCLASS1 16,113.0 100.0
Full
00001L5 LTOPOOL1 LTOCLASS1 20,840.4 100.0
Full
00001L6 LTOPOOL1 LTOCLASS1 19,088.1 100.0
Full
00001L8 LTOPOOL1 LTOCLASS1 800,000.0 0.6
Filling
E:\TSMDATA\SERVER1\DISK- DISKPOOL DISK 20,000.0 0.0
On-Line
1.DSM
There we see the VTL compression in action as well putting out something
that comes close to 1:2 compression on most volumes.
The range I wanted to use was 00001L3 -> 0999L3 but the VTL turned that
into 00001L3 -> 00001L8 :)
I can't really test performance because of the limitations in the
configuration I am given.
But even under small loads you can see on of the nice features of this
VTL come to life, the system automatically spreads it's load over the
disks so that when I run 4 streams to tape all disks receive about the
same amount of load, and this is balanced with every stream of data
coming into the VTL.
I can't say anything but that is works as expected.
If any of you would like to see something tested on a VTL like
this..maybe I can be of service :) !
Regards,
Stefan
|