I manage the West Coast portion of an international network for my company. Overall there are seven network managers in five countries. The company doesn't have an overall IS director. We, along with the application development people, all report to the controller. Like you'd expect, it is hard to get things done sometimes. Now I'm on a crusade to clean up the global network because it has developed in a very haphazard manner over the past few years. We have Windows NT, NetWare, Unix, and even some Banyan Systems Inc. systems. We have Cisco Systems Inc., Nortel Networks Corp., and 3Com Corp. There is no standardization at all.
What I'd like to do is get us entirely on TCP/IP, with a single vendor for the switches and routers that we use. Right now, if I have to cover for another network manager, it takes a while to figure out what systems they are using and how to use them. So far, there haven't been any major disasters because of this, but I think it is bound to happen sometime. Am I off my rocker? The other network managers think so, but I am sure that a simpler network will cost less and function better. Next, are there any tools available to help migrate from all sorts of equipment (configuration files, etc.) to a single vendor? And what can I do to convince the other network managers?
Brooks: Well, you've got a full plate -- and a messy one, at that. First, I'm pretty surprised that your situation could arise in today's technology-as-god world. Most companies today pay a huge amount of attention to technology decisions, and I'm shocked that any company with an international network spanning five countries doesn't have a CTO keeping everything in control. So even though you didn't ask, I'm going to recommend that you march into that controller's office and try to get some kind of central authority appointed to make technology decisions.
As for being off your rocker, I don't think so. It really depends on what the heterogeneity is costing you. If you have a ton of systems but they all work fine and talk together, you're better off replacing them one at a time during the natural upgrade cycle. If you're having frequent problems that raise staffing requirements, corrupt financial data, or whatever, you need to do something more quickly. The fundamental goal of getting on a single platform certainly makes sense.
There aren't any tools that I know of that I'd recommend for the migration path. It has long been my opinion that the Holy Grail of network management would be a software layer that abstracts all the underlying hardware, so you can con-figure your network from a central place and it translates that configuration into whatever commands each piece of equipment wants. That way, you could upgrade equipment either to a higher version or to a different vendor, and the network management software would just put out a different representation of that same configuration data.
Sounds nice, eh? Well, it doesn't exist. What you can do is use some kind of network diagramming package to get a snapshot of the network layer at least and put notations on for essential configuration data. It's ugly, but it will work.
I'm going to leave the convincing people part to Lori, because I've never been good at that myself.
Lori: I don't think you are off your rocker, but you are in the middle of an administrative nightmare. I would have to agree with Brooks that you need to address this problem with your controller. Make sure your boss is aware of what you and your team go through to keep things under control and running. Maybe the powers that be don't realize how much you have on your plate and what hurdles you face. No one who is sane would want to manage what you have without a large IT staff and good IT direction.
I suggest documenting what you have: the installed components as well as the staff available for maintaining this infrastructure, what tasks are routine, and what a typical workweek is like. You should also come up with some suggestions for reducing the hodgepodge you have in place and adopting some standards. If you lay things out clearly and underline how you will save time, which equates to saved dollars, you'll be speaking the executive's language. Streamlining your infrastructure will not only make your life more bearable but will save the company money in maintenance costs, and the staff will not have to be savvy in all areas.
Brooks: Test Center Rx is undergoing a couple of changes. With this column we're now appearing first online, and a week later we will show up in InfoWorld's networking bonus section. Our topics will be related strictly to networking from now on, but as you know, there's plenty of room for confusion in that field. We'll do our best to blow away the mists and make things clear.
I said "we" in the last paragraph, but perhaps I should be clearer, too. Test Center Rx is also undergoing a personnel change: This is my last week as a co-columnist. I will leave Rx in the eminently capable hands of Lori Mitchell, who will be joined by the more-than-savvy Kevin Railsback, the Test Center's West Coast technical director. (He's from Texas, but we don't mind.) I'll be moving on to focus on my job with InfoWorld.com. It's been a good couple of years for me, and thanks to all the readers who have written in with advice, kind words, and interesting flames.
Lori: It's been great working with Brooks this year on Rx questions; I wish him the best working on our dot-com team. I will miss his sage advice. For our upcoming columns on networking, please continue to send your questions, concerns, and topic ideas to firstname.lastname@example.org.
Brooks Talley is a senior business and technology architect for InfoWorld.com. Lori Mitchell is a senior analyst in the Test Center.