Projects, BAU and how IT all works really

I thought it would be interesting to have a series the ‘phased approach to IT’.

You see, everyone has a different set of priorities. They all indirectly converge, but they are all equally different. It’s part of a series of posts that will follow over the next few days. 

1. The first is with ‘Mike’ who’s an architect at an enterprise somewhere in Canary Wharf. He’s wanting to upgrade all the Windows 2000 servers to Windows 2008.
2. The second is Chris – who’s got a wide ranging set of priorities down in sunny Canary Wharf in his organization that as ever remains nameless.
3. The third is Johnny, our application support manager, his priorities are “make it faster, make it more reliable but do it cheaply, I’m not made of money”.
4. The last is with Arnold, a CIO who works in an enterprise in the city which you wont have heard off.

The idea is to illustrate some of the commonalitites, some of the concepts that I decide on which have unexpected consequences, as well as what issues there are in getting work done, in doing the support, the strategy or the architecture.

Before I introduce Mike in the next post, here are some of the basics that we assume people often just know, just understand about billing, about the way IT works in a large enterprise.  Of course it’s different everywhere but the same if you see what I mean. Some enteprises are technology rich, investment rich, others are technology rich at the lowest cost and others hover in between. Let us begin…

• BAU (or Support) = is business as usual support. This is the team that is dedicated to keeping the lights on, your server serving, your application online, your desktop working, doing whatever it takes to maintain the existing IT infrastructure. There will be some BAU projects, like upgrade firmware, move data centers, because they are changes to the existing infrastructure as the business or IT situations mandate.
• Projects = these are specific requests for services/hardware/time to deploy an application, a new web site, re-install windows on a new server etc.  This is billed directly to the business unit involved.

So if sales decide they want a new DL380 G6, sales get billed:

• The cost of the server
• The internal ‘per man day’ cost to rack, configure and deploy the server including any storage/os/network and project management time.
• The software license and monitoring costs

If however as Mike’s post illustrates, we’re talking replacing all Windows 2000 servers, we need multiple business line sign off, a shared project ‘bucket’ with specific costs assigned to ‘man days’ (allocating me from fixing/configuring to installing Windows, and hardware/software).

Every server will need Windows, it will need the monitoring, database or layered software purchased – and remember when upgrading an operating system, often you have to upgrade all the layered bits like the database etc.

This kind of project is more difficult to obtain sign off:

• It requires more business buy-in – it works, why are you bothering me?
• It represents significant capital and man day investment
  • Do you want Windows 2008 upgraded on all systems or deploy that new web application or infrastructure for a new business line/business opportunity.
• It represents significant activities in the data centers – moves/changes, racks/unracks/decommissions all of which makes emotional people further emotional
• It involves swing kit for those systems which need to be on all the time.

It involves possibly upsetting development teams and a re-alignment of standards/best practice – saying sorry:

• A) Your application needs re-coded to a Windows 2008 compatible development platform.
• B) The new operating system standards are now: service accounts only, shares locked down, no administrative rights, no console access etc.

It’s therefore one of those world ending requests. Not impossible, but as with anything easier in stages, in little bites than as one single transaction.

Declaring that you wish to replace 400 servers isn’t going to be popular, particularly if the cost of that server is being billed directly to the owner:

• What’s wrong with my DL380 G2, it works doesn’t it? 
• Can we try installing Windows 2008 on my DL360 and if it works, we’ll use that, will the 2x18GB drives (in a RAID 1+0 set) be enough for Windows and my SQL database?
• I don’t want Windows 2008, I’ll have to re-certify/validate my application
• Will we suddenly find all these ‘servers appearing from the woodwork which previously ‘weren’t in scope’, what was acceptable just 18 months ago, might not be now – how many fibre cards do you now need?
• How many applications need made high availability or need data replication or disaster recovery?

There are so many elements that you need to think about in just changing hardware or operating system.

This site uses Akismet to reduce spam. Learn how your comment data is processed.