So what is the driving force for this ?

1. Savings in power, cooling and physical space in
datacenters, fourteen 1U/pizzaboxes has approx. 50%
higher power consumption and need for cooling than 14
Blade’s in a BladeCenter and additionally takes double
the space of a BladeCenter enclosure, and that’s a fair
amount of money in a year, especially if you today is
in a spot where the datacenter has to be expanded in
one way or another.

However there are some drawbacks, which means that if
you utilize the BladeCenters higher density and smaller
“footprint” in the datacenter you may be forced to
redesign the airflow and power cabling to handle the
more dense environment.

2. Cost savings in hardware and installation, since Blade’s
in a BladeCenter shares common hardware you often find
that you will reach a break even in hardware costs
already with only 3-5 Blade’s in the BladeCenter
compared with traditional rackoptimized server boxes.

3. Cost savings in expansion, expanding the BladeCenter
with an additional Blade at some later point, is
definately more easy and cheaper than adding a serverbox
in a rack, all cabling and other infrastructure is
already there, only insert the new Blade and “clone” one
of your existing Blade’s and you’re up and running.
So even if a Blade is somewhat more expensive in
purchase it will alltogether be cheaper in the end.

4. Cost savings in management, BladeCenter’s form a very
homogeneous environment managed with common tools,
however there are some drawbacks in this area since
a BladeCenter normally (in larger enterprises) involves
several competencies (server, storage, network..) which
all are used to use their own selected tools to manage
their part of the enterprise and this can evolve to be
a problem if not approached correctly when introducing
BladeCenter’s in the organisation.

5. And why not run virtual server solutions on the
BladeCenter as well, even with a rather low
consolidation ratio f.ex. 3-4 servers/Blade this will
be a nearly ideal environment for all the functionality
the virtual server solution provides you with.
But as someone already pointed out, not all applications
are candidates to be hosted in a virtual environment.

6. Redundancy, BladeCenter’s is constructed with redundancy
in mind, however it could be a good idea to have a
second BladeCenter enclosure with Blade’s clustered
between the two enclosures, but again that depends on
SLA’s and other strategies that you might have in
your company.

Some last comments:

- Since Blade mounted options as f.ex. local disks
not can be “hot-swap” (without loosing significant space
in the BladeCenter), you should really consider to run
somekind of virtualization or cluster software in (or
between) your BladeCenter(s), if not the application
already is redundant in some other way.

- As you probably already is aware of there is a struggle
between different vendors on the market where I should
recommend selecting a vendor that supports the
“BladeCenter Alliance” which to my opinion aims to create
an open standard and where you should expect to be able
to use a wide variety (at least theoretically) of parts
from different suppliers in the selected BladeCenter.
Selecting an vendor that supports open standards will
probably lower your cost in the long run.

The other thing is the network connections. We are a Cisco shop so we looked at the Cisco switch modules, but it came out to over $600 a port where we pay about $90 on the Cisco switchs. So I would check for cost for port and make sure you have the density to support it.

Other than that they have been execlent performers. We have SQL, Citrix and standard tope servers on them.