How GM saved a billion dollars

General Motors Corp.’s (GM) IT group has undergone a thorough transformation in the last eight years, one that has seen the company knock more than a billion dollars off its annual IT budget of US$4 billion.

Tony Scott has helped drive that change over the last five years as CTO for GM’s Information Systems & Services organization. He reveals that behind the savings is a massive effort to consolidate — going from 7,000 legacy systems in 1996 to fewer than 3,000 today — and ideally he’d like to get that number down to about 1,000. The backdrop for all this is GM’s unique 100 per cent outsourcing arrangement, which both helps and hurts. Network World’s (U.S.) Jeff Caruso sat down with Scott to talk about how the transformation is going.

How long have you been 100 per cent outsourced?

Literally, since we bought Electronic Data Systems in the mid-’80s. GM bought EDS with the notion of EDS becoming not only the internal IT department for GM but also continuing its business on the outside. In ’96 we spun EDS off as a separate company but entered into this 10-year master services agreement, which essentially makes EDS our primary outsourcer but allows us under some complex rules to compete and bring in competition.

Do you have a lot of competing outsourcers?

We do, yeah. After this seven- or eight-year period we have brought in a significant number of competitors, but EDS (Electronic Data Systems Corp.) is still the dominant player.

Then you probably don’t have a lot of the fears that other IT people might have in terms of outsourcing part of their business, losing that control.

Actually, it’s been the opposite. When we spun EDS back out, it was, what info about GM’s information systems do we need to know, and how do we regain that knowledge from this 100 per cent outsourcer that we started in ’96? So it’s almost a flip of the question.

Are there some things that you would prefer to have more control over, or are there things that you miss?

It’s a difficult question. I worked in a lot of different IT departments, never one 100 per cent outsourced before. For a company of GM’s size and scale, the outsource model absolutely works. We don’t even think about reverting to a different model.

There are some things I do miss. We do everything by contract, so one of the effects of this model is that in some cases your opportunity to make changes is coincident with your opportunity to change the contract. On the good side, that creates a cadence of change and a cadence of opportunities that lets you constantly re-evaluate the technology and the suppliers and so on. On the flip side, if you want to do something mid-cycle, sometimes it can be more difficult. But net-net, it’s an advantage.

What’s your take on offshore outsourcing, a very controversial topic?

It’s a difficult question for me to answer given the global nature of what we do — when you sell cars and have plants in as many places as we do and therefore the IT operations in those countries….

Secondly, we’re 100 per cent outsourced, so we have EDS or IBM (Corp.) or HP (Hewlett-Packard Co.) or AT&T (Corp.) provide all of our services. We have less than 2,000 IT people that are GM employees. So they’re not replaced by anybody, either offshore or outsourced. But they work in Singapore; Russelsheim, Germany; Maidenhead, England; and Sydney and all kinds of different places. It’s a global, international company, and then we use suppliers that provide support and have presence in all of those places that we do business. So I just don’t think GM is in the model that a lot of other companies are in that regard.

I know GM is pursuing Web services. How far along is your Web services development?

It’s moving along at a cadence. We’re beyond the pilot, experiment stage, but like any large company it takes a while for any of these things to become completely mainstream in the bulk of the efforts that you’re doing. We still have system development projects in the deployment phases that were started before Web services were at all understood. And we’re not stopping any of those things just so we can switch them to Web services. But as we look at the architecture of the next generation of systems that we put in, they’re more and more reliant on what you’d call a service-oriented architecture.

Can you give me some examples of how you might be using Web services?

It’s really taking units of work that used to be embedded in systems over and over again — things like directory services, identity services, things that manage parts, people, cost, suppliers and so on — and componentizing them and making them available as a service in the infrastructure or in the application architecture. That’s the general direction that I see it showing up in GM.

Last year you were asked whether you were comfortable using Web services externally, and you said the main things you were worried about were security and the scalability of that security. Do you still have those concerns?

Still — but I think we’ve made a year’s worth of progress on those. So I think the (solution) is just starting to show up. For me it really becomes real when we start seeing support in products that we buy from our IT suppliers, where they’re willing to guarantee and sign up contractually for the security and the controls that one wants or the quality of service that we need.

Another thing you’ve said is that you have a single-vendor approach, like a chosen vendor in different areas. What are the benefits of doing that, and what might be some of the drawbacks?

It depends on the area, whether there’s a single vendor or two. We don’t always say there just has to be one. In fact, we generally prefer a model where we have a choice — but it’s not five choices, it’s one, two or at most three, in a given space. For us the benefits are being able to leverage our scale across the enterprise. In the database example, if I support five databases, it means I have to pay an outsource supplier not only to develop on five different database platforms or integrate across five different ones, but from an operations perspective I have to have trained database administrators and support people and maintenance contracts and all of that for five different ones.

So part of the systems reduction activity that we’ve engaged in is focused on simplifying the infrastructure, and really all of that cost reduction, that US$1 billion, has come out of the operational side of the budget. What it costs just to keep things running for support and maintenance of all the disparate stuff that we had.

A lot of that cost reduction came from server consolidation?

It’s getting rid of multiple databases and the attendant support that goes along with it — going from 15 small servers to two big ones.

That US$1 billion is a huge number. Is there anything out of that where you felt that you were giving up something? Or was it really just a matter of cutting out inefficiencies?

To be frank, some people would probably say they thought they gave things up initially. This is just the natural human reaction to the furniture being rearranged; it’s just not comfortable at first. But after an initial period, the pretty universal reaction would be, longer-term, the right thing to do and no long-term ill effects from any of that. And if you asked our business customer — the guy that’s not in IT and may have not touched it as directly except as an end user — they’d say they’ve seen dramatic benefits in terms of reliability and consistency of service. We’ve seen our performance numbers, availability numbers, reliability numbers go way up.

Is that something you measure?

Absolutely. It’s built into the outsource contracts that we have, their service-level agreements. We get money back from the suppliers when they don’t meet those levels.

You mentioned recently that you are going to Windows XP. Why make that move now?

We’re past the first service pack or two, which is the first test. But we’re increasingly a mobile world, and we like the support that you get for mobility in XP. We like the reliability that we’re starting to see. And then there are applications that are starting to be designed for XP that need a bigger memory model or faster processors. In the high-end CAD arena, for example, some applications just aren’t going to run on Windows 2000.

One other technology I wanted to throw out there is VoIP. Is that something that you’re considering doing?

We have it already. Our non-North America sites have really been the leader in this, so Latin America and Asia-Pacific in particular have jumped on the bandwagon for VoIP the earliest. This is probably a technology that’s going to move from the outside in.

Are you seeing savings?

Yeah. And it’s not just cost savings. Sometimes it’s just easier to implement than the old. As we build new facilities, it’s a pretty easy choice today to put in VoIP rather than wire up for the old legacy environment.

Related Download
3 reasons why Hyperconverged is the cost-efficient, simplified infrastructure for the modern data center Sponsor: Lenovo
3 reasons why Hyperconverged is the cost-efficient, simplified infrastructure for the modern data center
Find out how Hyperconverged systems can help you meet the challenges of the modern IT department. Click here to find out more.
Register Now