Bladewatch.com

I was having a conversation with Jim, a ‘common services’ manager based in a midsize financial organization in the city. I’d met him for lunch and was asking him what he was currently working on, and at that, I’ll let the story below unfold, the organization is looking at it’s file services issues and working out what to do next.

It all came as a result of an issue they’d been having with data. You see the organization had one file cluster, two Windows servers in a cluster with terrabytes of data shared over several drives, each up to 800GB, each drive corresponding to a business line or department.

The business had requested more storage, Jim’s response was that they needed to start deploying a new cluster with more storage, with failover and then deploy data de-duplication technologies to sit through the data an therefore archive/delete what wasn’t needed.

Below is the conversation we had, I’ve removed any customer/business specific information:

Is the data application or business line specific?

No it’s shared file data which has been hosted on our shared file cluster server, so the J drive might host IT and IT development data. Each share corresponds to a business line, but we don’t run applications off the file server, except for a few specific desktop functions but not in high volume. It’s static data.

Are there any existing quotas in place restricting data growth/size?

No we create a share on a drive and the business units on that drive share that space with whoever is also on that drive. If a share gets too big, we move that share on to its own separate drive.

What archiving do you have in place?

We periodically scan and delete old files, the data is also backed up to tape.

Ok, but what stops me hoarding data on the file server, keeping everything online?

Well nothing really, but we do delete old files when possible. Often the business say no though.

What are the backups like?

They’re ok, we’ve been having issues with capacity but that’s to be expected as the amount of data we have to backup increases.

What’s the lead time on a restore? If I log a call how long will it take?

Well, restores are not an incident, they are dealt with, but not as a priority, the priority is resolving incidents. It’s a lead time of a day or two.

Is there a per gb cost for putting data on the file server?

Our existing billing process is based around the per device cost, you pay a fee for your desktop or laptop, that includes a fixed fee which pays for your file and print. So no, there is no fee per gb, but then I don’t think the business would pay that if there was.

It’s an interesting coversation and highlights a few issues. IT and the business want to fix the problem, and they’re throwing money at it, they’ve been told four new servers plus fibre cards and a lot of storage. However, if we take a few steps back, before we spend money do we not need to fix our core? Fix the causal factors (as you get taught in history), not the secondary or tertiary factors?

The causal factors are:

• Lack of standards for data archiving
• Lack of data limits or quotas – I can backup my pc onto the file server
• The billing system does not encourage people to think about the data they are consuming
• The restore process is not responsive enough – could this be making people keep everything online?
• There is no separation in data duties between priority and back office data, IT might fill up the drive with restores, stopping development with their SourceSafe project.

The secondary issues factors are:

• File server is getting full
• Drives are getting fragmented and therefore slower at supplying data to the end user
• Backups are taking longer to run

Do we not in essence simply need to turn it around and have a discussion, a data amnesty if you like with the business sponsors and say, “we can either spend a few hundred thousand pounds in parts and people, or we can re-engineer our existing setup and fix the issue?”

• When we hear that the users don’t want to delete their data, are we asking the right people?
• When we say restores are not a priority, do we need to change that, either through giving users opportunities to restore data, or establishing a micro team of two or three people that only do restores, rather than bundling it under whatever department it currently sits?
• Have we thought about isolating user, business and private data rather than combining all three data forms on the same system?

When you ask Jim, the challenge he has is he knows this:

“I know, I know, we need to implement data archiving, we some kind of quota system, and we need to re-organize the way we do file services, but we’re at end game, I’m getting told by the business, we’ve got budget, just fix it, on the one hand, and a general understanding that I never get funding when it works. That’s easier to take the funding, buy a new system and migrate users than it is to try and fix the existing one and retrospectively install rules and procedures.

Often it can be easier to ‘start again’ blame the new system and say it’s best that way than to go back and start changing things with people saying that important phrase, but why, it works, leave it alone, what have you done.

Key, we don’t get investment or funding for shared IT services unless they’re broken and it might take months to sign off funding, so whenever we get the chance we take it.”

With that in mind, are we spending money we don’t need to spend in order to avoid an emotional conversation, to avoid some office politics? How much of the IT budget city wide are we ‘mis-investing’ simply because we are unable/unwilling to communicate with our business sponsors and end user community?

In essence by making funding a challenge, are we not increasing our costs, because although I could fix what we have, you wont give me money for future investment, how do we turn this around and develop trust and understanding?

No related posts.

Related posts brought to you by Yet Another Related Posts Plugin.