Can we define an infrastructure cost? First of all can we define ‘infrastructure’

It’s been a busy few weeks and for that I apologize (in terms of posts). I’ve been meeting with some colleagues, and participating in some interviews which I am due to post this week. In the meantime, I thought I’d relay this interesting conversation I had with Chris. Let me give you some back ground. In the enterprise the ownership of servers tends to be split between business lines, so server8 might be a PeopleSoft server for the HR team, the DC (domain controller) would be labeled as IT, and server19 (batch server) as business, in this case we’ll call it xtrad. At the same time, those servers which work for everyone, that provide a service, the email, the file servers, the domain controllers or patching servers will be labeled as IT, plus (and I mean this politely), anything else that no one wants to pay for, the file transfer box, it provides an invaluable function, but I can’t bill that to 14 different people, so IT ‘owns’ it, remember there typically isn’t a structured unit cost for things whether we’re talking pc or server. There is no, £5 a year for Windows, £5 a year to be in the domain, £5 a year to have a personal drive, £5 for security patching, £250 for the software support, £150 for hardware support and £100 for pc rental. There is a per unit cost, that is, a pc is say £900 a year for everything. You want a new pc, £900 please, whether it’s on the domain, whether you need software support, or a personal drive (it could be a mini test grid machine), you pay the support cost. With this in mind, Chris who’s still in Canary Wharf saving severs from emotional behaviour, called me to discuss an issue he’d had. Chris’ comments are in italics, I’ve done what I can to protect his identity, he still remains below the radar.

Have you ever had a re-labeling excercise?

What do you mean? A physical walk down stairs and check there are labels on the server?

No, not that,  something where you’ve got to reduce the number of servers IT owns?

Why what’s going on, do tell me more! Has someone done a data center audit or server inventory?

I’ve done it several times in the past, the best thing to do is go through it a few times, to see what servers you have, what they’re used for and go from there.

Yes we’re consolidating the data centers, and the latest report shows IT having 40% of the servers which sounds a bit wrong. So I’ve got to go through and verify the following details for each server, is there anything I should think about, what would you rule as IT?

• The server role/application
• The operating system type
• The server owner
• The server business line
• Server type
  

It’s a matter of debate I suppose. I like to break it down into ‘Infrastructure service’, and Infrastructure. So Infrastructure service would be a box that is owned by IT but provides a service to the end user, such as notes, DNS, WINS etc, where as infrastructure could be anything from a networks switch/server to a file server for the IT teams (repository for security patching/builds/software source etc). So:

• Infrastructure service - DNS, WINS, domain controllers, file transfer, middleware/monitoring (TNG/insight manager/IBM Director/HP OpenView, SMS, VMWare Virtual Center
• Infrastructure - network switches, file servers/IT specific systems like the server registry or inventory tools
• For example, DNS1 = DNS1, DNS, Windows 2000, Networks, IT, Infrastructure Service

It’s a topic of debate that is set to continue, and Chris managed to complete the report for his deadline, if you have any comments, do get in touch.