[EMAIL PROTECTED] wrote:
I routinely hit 100% CPU utilization on the Via C3 1GHz Mini-ITX
systems I use as backup servers. I will grant you that the C3 is not
the most efficient processor, but I am definitely CPU-limited. I too
have 512MB RAM, but the machines are not swapping. And that's with 1
backup at a time running (there is only one system to back up).
[...]
Like I said, these machines are not swapping during a backup. They're
just busy. Is this not normal? Is there a way to improve this?
Yes, do not use VIA C3 processors or motherboards. They are severely
underpowered at any clock rate. :-)
Seriously, I looked into using them for some dedicated processing
machines, and when I saw the benchmarks, ran screaming. They make fine
toy web servers, mp3 streamers or maybe low-end office use/browsing.
Chances are, they're spending most of their time calculating hashes,
where it is failing to perform. Like I said, I have a PPro 200 doing
two backups simultaneously, one machine of which takes about 25-30 hours
on a full backup to backup ~120gb over 802.11g *the first time only*,
but afterwards even full backups only take about 2 hours. The data is a
mix between many many small files (typical windows development box) and
a few large video files (video editing station). The GUI is slower
during this time, but not unusable. I am running mod_perl, though,
which I'm told matters a lot.
A system with only small files will have more work to do and more
packets to send, and more hashes to compute, so YMMV. If that is your
bottleneck, that is. I don't know the VIA chipset personally, so
perhaps it has real problems with memory throughput. If it doesn't have
a dedicated DMA controller, it will have to move memory using the CPU,
which will show up as usage the same as running a program would. Being
a mini-itx, I would suspect this.
It would be nice if there were a chart of people's backup system
configurations and their backup time performance over a specific network
type, just so users have some idea what to expect. The variability in
performance might make more sense with more data points.
Thanks,
JH
-------------------------------------------------------------------------
Take Surveys. Earn Cash. Influence the Future of IT
Join SourceForge.net's Techsay panel and you'll get the chance to share your
opinions on IT & business topics through brief surveys - and earn cash
http://www.techsay.com/default.php?page=join.php&p=sourceforge&CID=DEVDEV
_______________________________________________
BackupPC-users mailing list
BackupPC-users@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/backuppc-users
http://backuppc.sourceforge.net/
