Re: [rdiff-backup-users] rdiff-backup performance stats.

[dean gaudet]

On Tue, 23 May 2006, Steve Clement wrote:

> This is over a gigabit link SATA Raid 10 , conclude for yourselves.

i'm guessing you're saying the results are bad...

initial network backup is "well known" to be slow... in fact it's slow as 
soon as you use separate processes, as can be demonstrated with:

rdiff-backup --remote-schema '%s' src 'rdiff-backup --server'::dst

which doesn't even need a network... and you can see it on tmpfs too, so 
not even any disk i/o, and contrast it to the single proc case 
(rdiff-backup src dst)

if i had to guess it's due to the marshalling/unmarshalling code, but i'm 
just totally guessing.  patches welcome :)

i tend to hack around it by using rsync for initial backup then doing an 
"rdiff-backup --force" to turn the target into an rdiff-backup repo... 
then it's reasonable from then on out.

-dean