As I discussed last week, the benefits of virtualisation are
many: lower physical space requirements, reduced hardware costs, reduced power
consumption and cooling, and the ability to rapidly deploy new services. There
are, however, some increased risks which come with virtualisation. Lets take a
look at what they are and whether there are ways of neutralising them withoutlosing the benefits gained.
The stability of our virtual systems will, of course, rely
on the host operating system. If the host system crashes or is taken down by
some other problem (a virus, perhaps), then all of our guest systems on that
host will be taken down. This ties into an obvious risk behind migrating many
servers and services onto one physical platformthat a single hardware failure
will affect multiple virtual systems, putting all of our eggs in one basket, asit were.
Thankfully, some of the benefits of virtualisation can help
us overcome this risk and increase the resilience of our systems overall. How? First
of all, the nature of virtualisation means that backup procedures can be
greatly simplified; snapshots of complete virtual machines can be taken periodically
for very fast recovery. For those using hardware virtualisation, there is the
additional benefit of hardware isolation, meaning that recovery no longer needs
to be made to a host with an identical hardware configuration. In my opinion,
the greatest gain conceived by virtualising services and systems is the most
comprehensive safeguard against hardware failurethe increased affordability of
high availability computing. Sure, by consolidating four servers on to one
physical platform we have made substantial savings; however, we could choose to
set up a high availability cluster. Consolidating four servers into two will
still produce substantial savings and with increased redundancy over the single
server-single service model; we are not only negating the increased overall
risk posed by hardware failure, we are making very real gains as high availabilityredundancy is created.
One risk posed to smaller companies is not directly caused
by potential vulnerabilities or bugs. In an effort to maximise savings
(possibly egged on by the availability of free server virtualisation software),
the risk is that projects may be pushed forward into production without the
necessary investment in staff training or consultancy. This can result in
inadequate capacity planning, incomplete configuration of both the host and
guest operating systems, and a poor understanding of the administrative
requirements to keep things running smoothly after deployment. These
misjudgements could lead to poor performance and possible system failures. In a
production environment, this can result in spiralling costs through either loss
of revenue or consultant's fees (to fix the mess). Its therefore necessary to
make sure that the planning phase of any virtualisation project is
comprehensive and not fleeting; staff training should also be considered as anessential component and unavoidable cost.
As you can see, there are very real risks which need to be
taken into account when deciding whether virtualisation is the right choice for
your organisation. I personally believe that with the use of high-availability
computing and snapshot-backups, these risks are negated. Rather than it being
the case that virtualisation is increasing our vulnerability to hardware and
system failures, it is actually increasing
resilience while still saving resources. A very interesting reference on the
topic of high availability configuration of virtualised systems can be found here.
Have you recently undertaken a virtualisation project? What problems
have you come across? How did you overcome them? If you are still consideringmaking a move, what potential issues are holding you back?