On Fri, Feb 09, 2001 at 04:58:51PM -0800, Alfred Perlstein wrote:
> * Mike Castle <dalgoda@ix.netcom.com> [010209 16:57] wrote:
> > On Fri, Feb 09, 2001 at 04:23:10PM -0800, Alfred Perlstein wrote:
> > > Anyhow, if the idea is just to get a nice backup system, you could
> > > do a pg_dump and write the output to a NFS mounted FS, there's probably
> > > less that can go wrong with a large sequencial write than heavy shared
> > > read/write/seek.
> >
> > Still new to this but....
> >
> > Can pg_dump be piped through gzip et al? Use a few cpu strokes to save
> > network bandwidth may be beneficial.
>
> Yes it can. It's a cpu/space/network tradeoff, if you use low
> compression you might do pretty well, saving CPU and at the same
> time getting pretty good compression on plain-text.
I regularily use sth. like
/usr/bin/pg_dump dbname | ssh -C other.host /usr/bin/pgsql dbname
for this, "-C" tells ssh to compress data on-the-fly *and* your connection
is encrypted, allowing for doing this to a remote machine securely.
Installing your public key (identity.pub) on the target-computer in
~/.ssh/knownhosts will allow to get this running without a login dialog.
You may set your account-name on other.host together with enforced
compression and higher (or lower compression-levels) with an entry for
other.host in ~/.ssh/config on the source-host (well, man ssh ;-)). E.g.:
Host other.host
Compression yes
CompressionLevel 9 (1-9 available, 6 beeing default)
User zeibigm
ForwardX11 no
ssh uses an gzip-agorithm for this.
Best Regards
Mirko
--
mirko@picard.inka.de
http://sites.inka.de/picard/