BackupPC-users

Re: [BackupPC-users] Optimizing backupPC

2008-05-04 02:28:08
Subject: Re: [BackupPC-users] Optimizing backupPC
From: dan <dandenson AT gmail DOT com>
To: backuppc-users AT lists.sourceforge DOT net
Date: Sun, 4 May 2008 00:27:59 -0600
You should use Either raid1 or raid10.  avoid raid5.  the reason software raid is likely to be faster than hardware raid is that the software raid gets the benefit of your CPU, which is quite fast while many of the more affordable hardware raid cards have a weaker CPU.  Also, and maybe more importantly, is that linux software raid is completely controller and disk agnostic.  It doesnt really care if the controllers are different, or if some driver are added as whole drives and if some have raid partitions.  the slowest component determins performance though, so 1 slow hard disk and 5 fast ones will make 6 slow hard disks.

I routinely use the Backuppc_tarCreate to push off backups of systems to DVD.  I also mirror my whole backuppc server to its twin at another location, but be aware than you need a lot of RAM to rsync backuppc pools between two machines.  you need about 100Bytes per file, which adds up pretty quickly.  I don't bother making an additional pool backup, I just make archive disks for the host as I will never need to take advantage of the file pooling mechanism after my fullcount number is up(i do 3 full years of monthly backups) and then I will have long term DVD backups for each month kept in a safe.  DVD may not last forever, but how many people need more than the lifetime of a DVD kept in a safe?(rated for 25+ years)



On Sat, May 3, 2008 at 10:37 AM, Tino Schwarze <backuppc.lists AT tisc DOT de> wrote:
On Fri, May 02, 2008 at 04:40:04PM -0500, Chris Baker wrote:

> I'm considering a new backup server and am wondering what hardware I need
> for it. I would like to get one with two 100 Mb network cards so they can be
> teamed. (Gb cards are useless as all our switches are 100 Mb.) I'm also
> looking at getting some SATA drives and going with RAID 5.

Go for RAID10. BackupPC's accesses data very randomly because of the
hashing-pooling scheme. It also creates a lot of directories and small
files, so get very fast drives (with low seek times, throughput
shouldn't be an issue).

> I'm also going to divide the full backups into five groups. Some machines
> will have full backups on Monday, some on Tuesday, etc.
>
> BackupPC actually recommends software RAID over hardware RAID. Please let me
> know what you think.

Hm, the main reason for this might be that the OS is able to optimize
disk access better with SW-RAID. With HW-RAID, the OS only sees a huge
block device, it doesn't know of the individual disks nor stripe size
etc. of the RAID container. If you create a partition on a RAID5, you
might end up with the partition starting at the middle of a stripe. All
file system blocks will get "misaligned" and a lot more disk activity is
needed, especially for writing.

With SW-RAID, the OS knows about the hard disks and does the striping
etc. itself. So you may safely create a partition, add it to a SW RAID,
add it to LVM, create logical volumes and be assured that the logical
volume is nicely aligned with the RAID's stripe size.

My advice: Take your existing pool and perform some benchmarks! I'd like
to do it myself since I've got severe performance problems with my
server and I'm not yet sure what's the cause of all this. Unfortunately,
I don't have a spare backup server to play with. :-(

> Once our new backup server is in place and backing up everyone, I would also
> like to back up the BackupPC server to tape (probably LTO-4) or an external
> hard drive for off-site storage. I have also read that there could be issues
> with this. I am probably just going to do one full backup to media per week.
> I would also love to hear whatever recommendations anyone has for this.

I wrote myself a shell script which get's called by my Bacula job. It
creates TAR files for servers which haven't been saved to tape for some
time. Because of the heavy I/O and performance issues, it only manages
to tape-backup a few servers each day (about 3-5).

> I'm still wondering if such a backup plan is even possible. We have a lot of
> data.

Generally speaking: BackupPC's performance is mostly affected by the
number of files, not by their size! If you've got a lot of files and
directories, performance will drop.

HTH,

Tino.

--
„What we resist, persists." (Zen saying)

www.craniosacralzentrum.de
www.forteego.de

-------------------------------------------------------------------------
This SF.net email is sponsored by the 2008 JavaOne(SM) Conference
Don't miss this year's exciting event. There's still time to save $100.
Use priority code J8TL2D2.
http://ad.doubleclick.net/clk;198757673;13503038;p?http://java.sun.com/javaone
_______________________________________________
BackupPC-users mailing list
BackupPC-users AT lists.sourceforge DOT net
List:    https://lists.sourceforge.net/lists/listinfo/backuppc-users
Wiki:    http://backuppc.wiki.sourceforge.net
Project: http://backuppc.sourceforge.net/

-------------------------------------------------------------------------
This SF.net email is sponsored by the 2008 JavaOne(SM) Conference 
Don't miss this year's exciting event. There's still time to save $100. 
Use priority code J8TL2D2. 
http://ad.doubleclick.net/clk;198757673;13503038;p?http://java.sun.com/javaone
_______________________________________________
BackupPC-users mailing list
BackupPC-users AT lists.sourceforge DOT net
List:    https://lists.sourceforge.net/lists/listinfo/backuppc-users
Wiki:    http://backuppc.wiki.sourceforge.net
Project: http://backuppc.sourceforge.net/