Reply to Message

Contrary to other major vendors, we at Fujitsu-Siemens spend millions of dollars designing and testing various housings and components in different combinations. We are not only regulated by strikt government laws and market requirements, but more importantly, by customer input.

This article does touch on a few important subjects, being heat generation and related ventilation subjects.
However, who is the real audience for this story?

It can not be the major companies, as they all could not be bothered with these "trivial" subjects. Unfortunately for them this is also why they make many mistakes in their choices and spend far to much money due to these mistakes.

So I must assume the article is intended for small and medium companies.

When deciding what kind of server you want, might I suggest that the person responcible for this choice, first answers the following questions...

"What do I need the server for? File/Print serving, running a database, etc. What software will I need for this task (OS and Applications software)."

Knowing what you need, you can determine what specifications the required server needs to have.

This is where it becomes tricky and gets to a subject which is not covered by this article... Server sizing!

Does the target audience for this article have any idea how to size for instance an Exchange or SQL server?

So now a potential "server-owner/buyer" should have an idea what he/she needs.

Then the next question becomes most important...

"What will it cost my company, if the server we use is not available (for whatever reason)for 1 hour, 1 day or 1 week?"
Now this covers a number of costs like "loss of productiontime" (depending on howmany users are connected this can become a high cost in itself), cost for transport and repair or replacement and cost to your standing with your customers and this last one is a major point to think about!! This can cost you your business if your server fails!

These answers together and possibly a few more, form the basis of the level of "uptime" (time that the server has to be productive for you to run a business succesfully)required from your network in general and your server in particular.

You now need to decide on your server protection strategy....
Backup to tape....backup to disk....
Redundant powersupplies, redundant fans,
Harddisk raid strategy, etc. How about server insurance incase of fire or theft?
In case of defect...and you built your own server or bought a cheap "home made" one, can you still get the required components and at short notice? If you need to take the unit in for repairs, will you get a replacement unit to cover the repair time. If so, how much time and effort will it cost to re-install the temp.server so you are productive again?

There are so many issues concerning the choice for a server. The subjects concerning the chassis are just a very small part of the choice if at all important.

I can give you one example of a local customer where things went wrong...
The customer is a small transport company employing some 20 people and running I believe some 5 lorries (trucks).
They decided that a server is no different than any other computer and bought a pretty hefty "home-made" tower server.

The first problem they had was that they run dedicated transport/logistic software for which no really good sizing info was available and also they had no idea how much computer performance they needed overall.
They decided to send everybody home early on the friday afternoon so thay could start running the backup which would take 12 hours. The plan was to install the new server during the beginning of the weekend and use the sunday afternoon and evening to restore the data to the new server hardware. This meant they could only start testing everything on the monday morning during what are normal production hours.

Although at first all seemed well, on monday afternoon the server "crashed" and from then on kept "crashing" at different intervals for no obvious reason. The local supplier supplied new hardware but after loosing the whole of the monday's production time, the problem started reoccuring. On wednesday, we were called in and supplied one of our purpose designed and built servers. And on thursday morning the customer was up and running again. The difference in price between the two servers was in the region of some 10-14000 dollars, which is a lot of money, which ever way you look at it.
I later asked the company director what he estimated what the total cost had been for the downtime. He said he could not yet put a real figure on it, but it had meant that due to the total unavailability of the transport and logistic software and data, 10 people had been told to stay at home on full pay. The five trucks costing more than a house a piece were standing idly at various locations, some holding cargo and no where to go, cargo had been left at customer sites and had perished (they transport fruit and flowers) and this not only meant hefty financial penalties from customers, but also the fact that customers had transfered to other transport companies! So finally the director indicated that he could have given every person in the company the 14000 dollars difference in purchase price and bought the more expensive but working server, for far less money than they had lost now!!!

A very very expensive lesson to learn.

So again... a server chassis is the least of the troubles. Focus on your requirements and business needs and let us professionals focus on designing and configuring the server hardware to support you.

You will be able to sleep a lot better!