Information Technology Reference
In-Depth Information
the application, either manually or automatically. When that step is not enough, re-
starting the computer usually is, assuming it runs at all. But how do you know that
a restart is the best solution? Did the intruder change data, or even modify the ap-
plication? Did a software bug cause data corruption? Is all of the data recoverable?
In some situations, the only safe alternative is to restore data, or the whole
system, from backup tape. This effort can cause lengthy service outages and aban-
dons any legitimate transactions executed between the time of the most recent
backup and the time of disaster. Organizations with multiple data centers may use
storage replication to minimize the amount of lost data.
Some of these problems can be solved more easily if virtualization is in use. We
will explore some of those methods in the next section.
9.2.2.4 Disaster Recovery
As we discussed earlier, virtualization reduces the attachment between a work-
load and the hardware it is using. This looser connection makes it easier to think
of VEs as not being tied to a particular computer, which in turn simplifies the
planning for recovery and increases flexibility. Virtualization management soft-
ware can assist in the recovery of services and tracking of the physical location of
those services. The large number of VEs that a data center might include makes
virtualization management software a necessity.
Some virtualization solutions manage each VE as a single storage entity, per-
haps even storing it as a single file. Combine that capability with a file system
that can quickly take a snapshot, or with storage replication, and you have a new
method of restoring a service after a problem occurs. Not only is application data
replicated, but the application and OS, and their respective configuration infor-
mation, are also stored in a convenient package.
For example, the management platform (or storage subsystem) for a VE can
periodically take a snapshot of the file system in which the VE is stored. If the
application or VE fails, the workload's service can by restored by booting the VE
from the snapshot. The original copy of the VE remains, waiting to be analyzed. A
thorough forensic analysis should allow you to determine whether the problem was
caused by a person, accidentally or maliciously, or by a software or hardware error.
Copies of VEs should be kept on shared storage. This strategy allows you to
quickly restore service on another, sufficiently similar computer, even if the origi-
nal computer will not start at all. It also gives you the flexibility to restart the VE
elsewhere even if the computer can start, and to avoid using a faulty component
such as a bad memory module. If you start the VE elsewhere, you can replace
broken components, perhaps moving the VE back to its original home after com-
ponent replacement is complete.
Storage replication tools enable you to keep an identical copy of any data, in-
cluding data that represent VEs, synchronized across large distances. Using this
Search WWH ::
Custom Search