Backup & Recovery

 View Only

VDP reliability and job durations

  • 1.  VDP reliability and job durations

    Posted Oct 25, 2013 11:49 AM

    Hi there,

    whilst the backup utilities in the 3.x series and early 4.x series always used to run fine for us it's nothing but down right misery since then.

    VDP 5.0.x used to last on average for about 2 weeks. Then it went corrupt and needed to be replaced. Do note it did make backups of all machines.

    VDP 5.1.x more or less the same but it usually lasted for about 1 to 1.5 months.

    VDP 5.5.x is down right disappointing again. A new roll-out with only 30% of the machines in there already crashed after 3 days. With only 60% of that 30% actually backed up. It's running again but it frequently takes along the lines of 4.5 hours!! to take an incremental. I'm not sure what CBT is doing, but it appears it's not doing anything whatsoever... Did I mention that for many machines the first full backup actually went faster than the incrementals?...

    AppAssure does incrementals on many of the same machines, on the same hardware in less than 5 minutes. On some of the machines with a lot of changes it might actually be 10 minutes. It comes with a very different price tag tho' :smileyhappy:. That doesn't quite explain this extreme difference though.

    I have no clue where to start looking either. After only 3 days the absolutely fantastic logging (in fact there's so much logging I'm guessing that's what's keeping the machine busy >95% of the time... it even logs stuff about VM's nearly every minute when it's *NOT* in the backup window...) has 1.5GB in text files. And of course not in 1 text file, no... they're scattered all over the place, there's many dozens of em.

    Unfortunately that's a bit harsh to handle, by the time I'm done reading all of that we're probably at VDP 29.5, or more probable new product 2109 since they seem to replace the product with another every major release, or at least they did for 3.x, 4.x and 5.x. Hopefully they'll hire AppAssure for the v6 then because their stuff seems to be a factor 2 billion more sane.

    The amount of reading it does from it's own storage during backups is also insane. At least, I'd expect the maintenance window to be for that...

    Anyways, I'm mainly wondering 3 things:

    a) Is anyone actually using this product successfully? I sure hope it's my environment (mainly because if it's not I should draw some serious conclusions about VMware's credibility)...

    b) Anyone have any ideas what logs to look first? They're not only gigantic, there's also hundreds of em spread all over the place.

    c) Could it have to do with our lack of AD? We're just a small time company, it's used for hosting so definitely want to keep it separate from our own AD and since there's only 2 people managing it I really don't get why an AD would be mandatory. Everything else seems to work just fine...

    With 5.5 there's a performance analysis thingy, it reports:

    Performance ReportAchievedExpected
    Minimal write throughput (MB/s)52.923630.0
    Minimal read throughput (MB/s)111.429960.0
    Minimal seek throughput (seeks/s)4356.68320.0
    Total write throughput (MB/s)53.243330.0
    Total read throughput (MB/s)111.98560.0
    Total seek throughput (seeks/s)4451.11400.0

    It says passed...