Many, many B2D files in BE 2010 (13.0)

[birchanger]

I find myself in charge during a colleagues vacation and have been having problems with failed backups. I'm battling my way through these but am surprised by the time some of them are taking (e.g. 40 hours for 600GB for a B2D job).

I've noticed that there are many (hundreds) of B2D files, many of them small (under a gig). The result is that the B2D folder is always full, and chunks of anything from a few GB to under 1G are being overwritten when needed for the backups. At least that what looks as if it's happening to me.

Settings:

Maximum size for backup-to-disk files: 20GB

Maximum number of backup sets per backup-to-disk file: 100

Low space threshold: 1GB

Eack backup job is, in effect, one (normally virtual) machine, with data on the SAN-hosted D: drive.

I expect that a defrag of the physical disk might be useful, but isn't this multiple-file checkerboarding slowing me down? If so, how do I fix it?

[Kiran_Bandi]

What is your jobs configured to do when those start? Choose overwrite only

What is your media protection level set to?  Set it to FULL

What is your Media set configuration? Choose proper OPP and AP settings depending on your requirement.

All these things will effect B2D folder/files (Media) management.

Refer: https://www-secure.symantec.com/connect/blogs/faq-how-configure-media-setsb2d-jobs-and-backup-exec-proper-management-disk-space-while-using-

Regards....

[birchanger]

Rob,

Thanks for your help. It sounds as if the situation here is not unusual, although perhaps slightly extreme. I'll look at your suggestions for managing the B2D folder.

Kiran,

I'll take another look at the settings and report back.

[birchanger]

One thing I didn't reply to on Rob's post:

I assume that the BKF files are in a media set that has an overwrite date set on them? If so, then they should be automatically re-used

Yes, they do have an overwrite date, and are reused. But it seems to me that they retain their size after reuse. Maybe that's expected behaviour if there's no unallocated space? After all, in those circumatnaces a file can't grow.

I think I need a B2D folder clear-out. The school summer holiday starts after next week so I have six weeks to tidy this up, and I think it's just jumped to near the top of my priority list.

[robnicholson]

Yes, they do have an overwrite date, and are reused. But it seems to me that they retain their size after reuse. Maybe that's expected behaviour if there's no unallocated space? After all, in those circumatnaces a file can't grow.

Not sure - never watched a BKF closely in re-use. I don't know whether BE creates a new BKF using the same file name (in which case it would grow up to 20GB) or simply writes new blocks over the top.

On our live BE system, the BKF files are only 1GB each so it's less of a problem. The new BE media server I've just built defaulted to 4GB (Symantec must have decided 1GB was too small). However, somebody must have changed the default on your system to 20GB. That would be a good test of whether BE re-creates or simple re-writes BKF files. Reduce the size down to 10GB and see if your BKF files start shrinking to that size as they are re-used.

I think I need a B2D folder clear-out. The school summer holiday starts after next week so I have six weeks to tidy this up, and I think it's just jumped to near the top of my priority list.

Sounds like a plan.

Cheers, Rob.

[robnicholson]

I find myself in charge during a colleagues vacation and have been having problems with failed backups. I'm battling my way through these but am surprised by the time some of them are taking (e.g. 40 hours for 600GB for a B2D job).

I'm just looking at a new install of BE and also wondering why B2D is so slow. If my math is right, 600GB is 40 hours is ~256MB/min which is almost exactly what we're getting (250MB/min) so I suspect there is nothing wrong with your hardware set-up. This test is currently to an external USB 2.0 hard disk (it's just a test!) which according to ATTO Disk Benchmark can do 1800MB/min (30MB/s) so quite why BE can only manage a throughput of ~7 times slower is perplexing. We've also done tests to the local SAS drives and the backup speed is much the same so one infers it's not the speed of the disk system that is slowing things down here. Backup to LTO3 tape on the same system manages ~700MB/min which is x3 faster than disk. Something doesn't add up here and my suspicion is the BE backup-to-disk system itself - either on the server or the agent.

Failed backups, battling and surprise are words I can relate to

I've noticed that there are many (hundreds) of B2D files, many of them small (under a gig). The result is that the B2D folder is always full, and chunks of anything from a few GB to under 1G are being overwritten when needed for the backups. At least that what looks as if it's happening to me.

Your settings below suggest they should be all 20GB big except the current one. Ours are 4GB and most of the BKF files are indeed 4GB but there are ~10 which are less than that so we see the same thing.

I expect that a defrag of the physical disk might be useful, but isn't this multiple-file checkerboarding slowing me down? If so, how do I fix it?

I assume that the BKF files are in a media set that has an overwrite date set on them? If so, then they should be automatically re-used. Our current B2D system often fills up (too much data to backup, too small a B2D drive) so we do one of two things:

1. Sort the media list by allocated data and move some old ones into scratch media
2. More draconian (and you're on your own here!), manually delete BKF files over a certain age, re-inventory the B2D folder, find which media is in the catalog but doesn't have a "Media location", move those into "Retired media" and then delete them from retired media

But in general, your BKF files should all be 20GB except the one currently in use.

Cheers, Rob.

[Ken_Putnam]

Reduce the size down to 10GB and see if your BKF files start shrinking to that size as they are re-used.

I've never tried this, but am fairly certain that would only affect files created from that time forward 

The OP would have to move the BKFs to retired media and then delete then from within BackupExec and let BE recreate files as needed to get the average size down