Joey Kelly wrote:
>> Any other solution ideas? Maybe rsync for windows via SSH? Anyone set
>> anything like that up?
>>
>>
>
>
> Ok, call me cheap and set in my ways, but I'm going to be doing exactly that
> next week. I will install cygwin on the windows PC, and make sshd run as a
> service. Once I have that in place, I'm going to pull updates to a linux
> server using rsync.
>
>
It all depends on your objectives and the recovery point you want to
support. If you can do replication at the block level, perhaps
asynchronously, you can have your recovery point on the DR host within
seconds of the source. I've found a couple of non-commercial Linux
solutions to do this -- the "network block device" (
http://www.it.uc3m.es/ptb/nbd/ )which is a kernel module and an
application from 2004 called Pratima
(http://www.linuxjournal.com/article.php?sid=7265 ).
Of course, having your recovery point within seconds of the source is
only part of the solution. It also depends on the application you are
dealing with. An oracle database, for example, may not be of much use
if it's not put into "online backup" mode or completely shut down -- the
database won't be restartable. In other words, I'm assuming there's a
controlled disaster such as a hurricane entering the gulf. you
replicate continuously at the block level and then shut down normally in
that kind of scenario so you have a restartable application at the
remote site. But if something catastrophic and sudden happens you may
have more of a challenge to get back to some "sane" point to start your
app. In the oracle scenario I mentioned, this is typically achieved
through the use of Oracle archive logs.
If you had a mysql db, you'd probably be best off replicating within
mysql itself.
The point I'm making is that you have a lot of things to consider. More
often than not, rsync over ssh is a more than "good enough" solution,
but you need to understand just what that gets you in terms of
recoverability. If you don't have too large a data load and you can get
the downtime to send over a "known good" copy of your data then it
certainly fits the bill. And if it's just plain file data you want to
get off site and you don't care about not be able to replicate open
files, then you probably don't even need the downtime. But if you have
more stringent requirements or a more complex scenario, you may need to
go with another solution.
Getting the data offsite is not the goal. Getting it offsite and being
able to _use_ it offsite is the goal.
I definitely recommend Veritas Volume Replicator as a commercial
supported solution for this purpose. I haven't run it on Linux but do on
Solaris and I know our Wintel team does as well. Obviously, it's not
cheap.
___________________
Nolug mailing list
nolug@nolug.org
Received on 07/09/07
This archive was generated by hypermail 2.2.0 : 12/19/08 EST