is TSM really the best?

[rwhtmv]

So the boss now says to evaluate ANNUALLY all of what we use, and provide insight to other products, and why we do NOT use them. I just love micro-management. However, there are a few items with TSM that I have always wondered about. Please address these if you have a way to resolve them, or have found a workaround.

1. When a node is backing up and I have to CANCEL SE for whatever reason: how to know (other than file property inspection or dsmsched.log browsing and guessing) what is left to backup from that node that was missed due to the cancelled session. Is there a way to view the journal logs, or extract the information from the client, since it pre-scanned the filespace and knows what it intends to capture?

2. What is the best reporting tool/dashboard you have used? Most of the built-in stuff is just not good enough. Something to keep the boss happy with pretty information he doesn't understand.

3. Why is this true?: Node connected to T1 circuit with nothing using it except TSM client session doing a backup. No noise on the line, can move data via FTP at good speed. TSM server has only 1 session, and no procs running, quad core proc. Writes backup to tier 1 disk with plenty of performance to avoid bottlenecks. (Migrations zoom from disk to tape.) So....start a backup from ANY version of TSM client (tried them all). Client sends the files at nothing close to the speed it should. Node is also quad core, with no performance hit. I'm not talking about the aggregate speed, either, just plain old backup speed. Please refrain yourself from saying "Defrag". That is not it.

 

[moon-buddy]

1. When a node is backing up and I have to CANCEL SE for whatever reason: how to know (other than file property inspection or dsmsched.log browsing and guessing) what is left to backup from that node that was missed due to the cancelled session. Is there a way to view the journal logs, or extract the information from the client, since it pre-scanned the filespace and knows what it intends to capture?

Look at the dsmsched.log to determine what was transferred/backed up.

2. What is the best reporting tool/dashboard you have used? Most of the built-in stuff is just not good enough. Something to keep the boss happy with pretty information he doesn't understand.

I create my own - sorry, can't help you here.

3. Why is this true?: Node connected to T1 circuit with nothing using it except TSM client session doing a backup. No noise on the line, can move data via FTP at good speed. TSM server has only 1 session, and no procs running, quad core proc. Writes backup to tier 1 disk with plenty of performance to avoid bottlenecks. (Migrations zoom from disk to tape.) So....start a backup from ANY version of TSM client (tried them all). Client sends the files at nothing close to the speed it should. Node is also quad core, with no performance hit. I'm not talking about the aggregate speed, either, just plain old backup speed. Please refrain yourself from saying "Defrag". That is not it.

What is your resourceutilization setting? If this not set, try setting it to a value like 5 and see if the transfer has improved.

And YES, TSM is the best that I ever had used.

 

[Geirr]

Hi,
In my opinion - yes, TSM is the best. Not to say it does not have flaws, but so does every backup software package I've seen. My few cents worth on your questions:
1. There is no way of knowing. Can you do that with other backup software?

2. Agreed - I don't like Admin Center too much. There are a few choices that are good - TSM Manager and Servergraph being 2 of the most widely used.

3. Sorry, but my best guess is network issues with NIC, Switch, routers, and FW settings to look at. Also make sure the FTP test is the same (kind of) data as the backup test, and follow the same route. The problem could also be the source filesystem, not being able to keep up, for any number of reasons.

/G

 

[rwhtmv]

So I'm seeing that there is no possible way to know what was INTENDED to backup in a client that did not complete. DSMSCHED.log is not an accurate or efficient way to do this, by guesssing and comparing to archive bit setting for specific files. See my point is: TSM client KNOWS what to backup, because it already looked during the initial scan. Note to TSM architects: JUST TELL US where to find that information, so we can do our jobs better.

 

[fuzzballer]

I am in total agreement that TSM is the best. I have used HP Data Protector, Veritas and Microsoft DPM. None are even in the same category as TSM.
#2 - Take a look at TSM Studio, TSM Console and TSM Scripting by Spirit software. The second 2 are free and you can get a 30 day full working copy of Studio. This is IBM approved software for use with TSM
#1 & 3 - I think all has been said regarding these.

 

[JesperM]

Hi

1 - TSM can not report which files were not backed up. Such a list would in some cases contain millions of entries. However - when you start the backup next time, TSM will not spend time and bandwidth on files that are already transferred, and will in most cases continue more or less where it stopped the last time.

2 - I invite you to take a look at our cloud-based TSM reporting solution - Wizards Storage Portal. It will monitor your backup system 24x7 and also provide you with lots of graphs and reports to impress your Boss. You can even configure it to send him an SMS each time his PC has been backed up by TSM ;-) It requires no extra hardware and is deployed in minuttes.

3 - The communication parameters are important when backing up over thin lines. I would like to see some statistics from your system and the dsm.opt file. And we can easily figure out what is going on.

I have been working with ADSM/TSM since 1995 - and I have not meet any other system that provides the same robustness and scalability as TSM. Surely - you can find other products that will beat TSM on some areas - but the overall picture remains. Nothing beats a well-driven TSM system.