ADSM-L

Re: [ADSM-L] TSM for VE sizing vs standard incremental forever

2013-05-16 00:59:32
Subject: Re: [ADSM-L] TSM for VE sizing vs standard incremental forever
From: "Prather, Wanda" <Wanda.Prather AT ICFI DOT COM>
To: ADSM-L AT VM.MARIST DOT EDU
Date: Thu, 16 May 2013 04:57:33 +0000
Below are stats from a real-world run:  TSM/VE 6.4, incremental-forever.

All were incrementals in this run (usually we have at least 1 new VM, but not 
today.)
86 VM's (mixed 70/30 Windows/Linux)
11.29 TB of VM disks
 543.12 GB of data moved  (no client-side dedup)
Windows VM for the single data mover, settings:
vmmaxparallel  8
vmlimitperhost  2

That being said, I'm not sure you can come up with a universal "per TB" 
estimate.
It's going to vary depending on your local change rate, and especially how much 
of your data is DB backups still done with TDP's rather than the VM disks done 
with VE.

It's good you have experience at that site with TSM before VE; you can use that 
daily-change-rate as a starting point.
The ONE thing you can be sure of, I think:
The data moved using block-level incremental forever will be less than what you 
were getting with file-level incremental forever.

For non-Oracle, non-Exchange systems, if I have nothing else to go on for 
planning, I assume 10% change rate for traditional TSM incremental forever, and 
5-6% for block-level incremental (with no client-side dedup).  YMMV.

You noted that your back end will be all tape; be sure you read the 
"requirements for tape" section in the VE book.  The metadata has to be split 
out and stay on disk.

05/15/2013 21:47:56 Aggregate statistics for Backup VM command.
05/15/2013 21:47:56 Total number of objects inspected:           86
05/15/2013 21:47:56 Total number of objects backed up:           86
05/15/2013 21:47:56 Total number of objects updated:              0
05/15/2013 21:47:56 Total number of objects rebound:              0
05/15/2013 21:47:56 Total number of objects deleted:              0
05/15/2013 21:47:56 Total number of objects expired:              0
05/15/2013 21:47:56 Total number of objects failed:               0
05/15/2013 21:47:56 Total number of subfile objects:              0
05/15/2013 21:47:56 Total number of bytes inspected:          11.29 TB
05/15/2013 21:47:56 Total number of bytes transferred:       543.12 GB
05/15/2013 21:47:56 Data transfer time:                    6,680.61 sec
05/15/2013 21:47:56 Network data transfer rate:           85,034.98 KB/sec
05/15/2013 21:47:56 Aggregate data transfer rate:         87,833.83 KB/sec
05/15/2013 21:47:56 Objects compressed by:                        0%
05/15/2013 21:47:56 Total data reduction ratio:               95.32%
05/15/2013 21:47:56 Subfile objects reduced by:                   0%
05/15/2013 21:47:56 Elapsed processing time:               01:47:47
05/15/2013 21:47:56 Backup VM command complete
05/15/2013 21:47:56 Total number of virtual machines backed up successfully: 86



-----Original Message-----
From: ADSM: Dist Stor Manager [mailto:ADSM-L AT VM.MARIST DOT EDU] On Behalf Of 
Steven Harris
Sent: Wednesday, May 15, 2013 5:20 AM
To: ADSM-L AT VM.MARIST DOT EDU
Subject: [ADSM-L] TSM for VE sizing vs standard incremental forever

Hi All

I've been asked to do some ballpark sizing for a major rework of a TSM 
environment. Clients are linux and Windows under Vmware

This customer has been using standard BA client backups up until now, ie 
incremental forever, and the new env will be TSM for VE with the backend being 
tape rather than disk so no dedup possible.

Given that TSM for VE uses some funky differential backup technology on the 
changed blocks, how much data does your typical VE installation backup per day 
per protected TB?  Anyone with real world data out there?  Sorry it is such an 
open question but anything you can give me is better than a stab inthe dark.

Regards

Steve

Steven Harris
TSM Admin
Canberra Australia