[Gllug] MS & LSE ROFL!

[James Hawtin]

On Fri, Nov 27, 2009 at 09:39:32PM +0000, Bruce Richardson wrote:
> On Sat, Nov 21, 2009 at 03:38:03PM +0000, general_email wrote:
> > 
> > The T-mobile sidekick fiasco was due to an MS powered SAN upgrade
> > failure. It seems like it's not uncommon for large SANs not to be backed
> > up these days!
> 
> There's no excuse for that.  Databases offer replication facilities, SAN
> storage arrays have mirroring and snapshot features.  One of the most
> coommon reasons, though, why large SANs are not always backed up is that
> they are being used (quite wrongly) to support massive, monolithic
> filesystems which are so huge that they can't be backed up coherently.
> There's *absolutely* no excuse for that - very poor application design
> if you require a single filesystem to support it.  No need, when there
> are plenty of distributed storage systems to choose from (or, if you
> needs are uncomplicated, use a simple web-based service).
> 

Most backup software can use multiple drives to backup the same filesystem,
using one huge filesystem isn't much better to backup than multiple ones if
the data needs to be consistant over all of them at the same time. The san I
used to run a Hitachi Tagma Store, can a feature called shadow image on it,
wher the san could snapshot everything and you can then dump the image.

Tape Drive throughput can be solved by modern tape systems than provide
virtual drives to computers and backup the data to hard drives then write
that to tape drives later. 

What is true is most people are not will to invest money is buying a backup
solution suitable for there needs, and don't understand why its much more
expensive than the hard drives... 

I have an fibre channel LTO4 drive my computer cannot write data to from the
hard drive fast enough!

James 
-- 
Gllug mailing list  -  Gllug at gllug.org.uk
http://lists.gllug.org.uk/mailman/listinfo/gllug