Send your Storage question to David Hill today! | See other Storage tips



What seems to be the problem? IT organizations may confuse the terms "high
availability" and "disaster recovery." You need both, but if
you think that the two are synonymous you may do one or the other and think
that you have done the other. And that could be a problem as the two, while
interrelated, are not the same.

What do you need to know? High availability refers to the ability of an end
user to have a chosen application (and its associated data) available for use.
Recovery time objective (RTO) is a measure of that availability. High availability
is an RTO that is typically measured in terms of seconds or minutes of downtime
per year. High availability should be operational recovery that is transparent
to the user. For example, recovering from the physical failure of a single disk
in a RAID 5 disk array should be transparent to the user (except perhaps for
some hopefully acceptable performance degradation during the RAID group rebuild
process). Thus high availability is about not impacting the user application
experience day after day after day.

Disaster recovery refers to the ability of an organization to return applications
to normal operations after a disaster. A disaster is an event that requires
failover from a local to a remote site for an extended period of time.

Well, wouldn't you like high availability for a disaster recovery scenario?
That depends. If you are running a Web business, a quick failover to a remote
site may be necessary to keep your business running as your customers could
be anywhere in the world. But, if your business is confined to a specific geography
and that geography has been impacted by a natural disaster, such as a hurricane
or earthquake, your end user may not be able to use an application even if it
were available. The focus then is first on data preservation and then restoring
applications at a time that coincides with when end users can actually make
productive use of them.

Think of high availability for applications then as typically responding to
operational failures for one or a limited number of applications while disaster
recovery is a response to a systematic failure of multiple applications simultaneously
due to a disaster.

But what about two failures in a single parity RAID group or database corruption?
Either event could force a failover to a remote site. Isn't that a disaster?
The answer is no. Now having the remote disaster site and being able to use
it for a number of non-disaster purposes is useful and makes better use of the
assets at the remote site. However, the acid test is whether or not the equipment
at the remote site could have been located at the local site. If recovery would
still be possible, then it is not a disaster. In the case of a true disaster,
the equipment at the remote site could not be located at the local site and
still have recovery occur.

What can you do about it? My previous storage tips have touched upon this subject,
but some organizations still equate high availability with disaster recovery.
You need a combination of both, but having one but not the other does not excuse
you from the problems that might result. Now some call the combination of the
two continuous availability and that is useful. Where possible, end users should
never be impacted by problems that IT experiences. But to achieve that goal
you have to plan carefully to have the right hardware and software resources
to ensure both high availability and disaster recovery.

» posted by jnaze

Mesabi Group

• Email
• Print
• Share
  • Slashdot
  • Digg
  • Stumble
  • Reddit
  • Newsvine