Debunking the terabyte per admin metric
When I meet with high-level managers, one of the first questions usually put to me is "How many people do I need?" This is usually followed by someone saying that a standard for the number of terabytes managed per administrator (TB/admin) would be helpful. Sadly, this metric is meaningless. So, let's knock it down once and for all.

Let's start with complexity. If an administrator can manage, say, 10TB, does that mean he could manage one hundred 100GB arrays? How about one from each vendor? Complexity is the single most important factor in determining the number of administrators needed. The number and variety of arrays, different host platforms and host bus adapters (HBAs) and different management applications all have a huge impact.

There's also no indication of effectiveness over time. I bet I could manage 1,000TB for one day. Or you could fire everyone in IT today, and everything would probably still be working tomorrow. And think of the money the company would save! In my experience, most IT shops are understaffed and expected to be responsible for unrealistic requirements.

Focusing on TB/admin ignores other skill sets found in good management teams--operations, cost accounting, business analysis, management and the like. There's more to storage management than just managing storage systems. Someone must negotiate SLAs and calculate the financial impact of their choices.

Also, there really is no defined best practice for TB/admin. I found analysts suggesting 0.5, 1.25 or 1.5, while storage product vendors said five, seven or 10. The number has changed over time, too, from the same sources. No wonder everyone is so confused. When I see someone tossing around this metric, I smell a quick-fix claim or a scapegoat. The funny thing is that because most businesses run IT pretty lean, a simple calculation will show that their TB/admin beats the analysts and vendors claims.

So what do you really ...

How many administrators you need has to be based on skill sets, coverage and workload:

1. Count your skill sets. Include all your technologies--storage arrays, switches, host operating systems, backup products, storage management applications.
2. Divide by 1.5. Each person is good for about three skill sets, and each skill set needs at least two people, so that gives us 1.5. It's true--no matter how smart you are, you really can't know enough about more than three things to do a really good job managing them. And everyone gets sick, or quits.
3. Multiply by the size of your environment. I admit it, this one is a cop-out. Maybe you have 50 data centers, or one huge center with 50 storage systems. Whichever it is, you probably need more than just a few administrators. Here's a rule of thumb--a single administrator can manage about half a dozen similar storage systems and their associated connectivity and clients. So if you have 50 IBM Sharks, you'll need eight or nine administrators.

Does this sound like many people? A similar real-world business is likely to have three or four administrators. But they would probably be frustrated, reactive and generally not doing the kind of good job they'd like to be doing. And this real-world business probably doesn't have a business analyst to make sure that their customer expectations are being met, either. This is why I said most IT shops are understaffed.

Another option is reducing the complexity of the environment. Standardize on a single switch and block storage array, lose one of the host platforms and you only need four or five administrators.

If your business has decided to build a dedicated storage management group, they're on the right track. The challenge is to make sure that the group is well-defined and correctly sized to meet expectations. A group that's too small will be unable to meet user expectations, and will threaten their faith in the internal service provider concept. It could create an unstable environment, putting the business at risk.

< PREV PAGE   |   1  |   2  |   NEXT PAGE  >