Les Mikesell <lesmikesell@gmail.com> wrote on 09/18/2012 12:42:26 PM:

> On Tue, Sep 18, 2012 at 10:24 AM, Timothy J Massey
> <tmassey@obscorp.com> wrote:
> >
> > Fortunately, BackupPC is a "backup of the backup" right now, and is not
> > expected to be used for real.  Yet.  That's why I can take the time and try
> > to actually solve the problem, rather than apply band-aids.
> >
> > But that will likely end in November, if not sooner.
>
> It is more than a band-aid to have a warm-spare disk ready to pop in
> instead of waiting for a 3TB restore even with reasonable performance.
>  Be sure everyone knows what they are losing.


That is a good point, but if I ever have to do a full 3TB restore from BackupPC, the 12 hours a (properly performing) BackupPC will take is not my biggest issue.  I don't look at BackupPC as my bare-metal disaster recovery plan--or, at least, not my first line of such defense.  That's what snapshots and virtualization provides.  BackupPC is for file-level restores, and the odds of having to do a full restore from BackupPC is small.

> > I'm not.  I've got two of these new boxes built.  In both cases, they have
> > 2-4% wait time when doing a backup.  One is a RAID-5 and one is a RAID-6.
>
> Can you test one as RAID10?  Or something that doesn't make the disks
> wait for each other and likely count the time against the CPU?


Unfortunately, both of these boxes are in production, so they can't be reconfigured;  and I don't have enough parts for another one just yet.  It is something worth trying.  I predict that this won't make that much of a difference:  I've tested small-write performance differences between single-disk, RAID 1, RAID 10 and RAID 5 (but not RAID 6) before, and the penalties, while very real, were also very manageable.

But I'll see if I can try it again.

> > Might it have something to do with md?  Could the time that would normally
> > be considered wait time for BackupPC be counted as CPU time for md?  That
> > doesn't seem logical to me, but I can say that there just isn't any wait
> > time on these systems.
>
> Not sure, but I am sure that raid5/6 is a bad fit for backuppc
> although good for capacity.


And frankly, capacity is what I need more, with a certain minimum amount of performance.  I do not need top-performance, and am perfectly willing to sacrifice performance for capacity, as long as I could get, say, 50MB/s of BackupPC throughput.

50MB/s performance for a RAID-5/6 array should not be difficult, even with read/modify/write and small transactions.  I thought I had tested this workload on this array successfully, giving me more like 80-100MB/s on synthetic benchmarks.

But again, I'll see what testing I can do.

> Are you sure the target has no
> other activity happening during the backup?


I am sure they *are* seeing other activity:  they're file servers, mail server, etc.  But their loads are all very low across the board.

Timothy J. Massey

 
Out of the Box Solutions, Inc.
Creative IT Solutions Made Simple!

http://www.OutOfTheBoxSolutions.com
tmassey@obscorp.com
      22108 Harper Ave.
St. Clair Shores, MI 48080
Office: (800)750-4OBS (4627)
Cell: (586)945-8796