Government Operations Committee on Feb. 13th, 2012

[Ron Cannan, Conservative, Kelowna—Lake Country, BC]

Ultimately, the taxpayers are our customers. We're in the business of serving Canadians, so we want to deliver the services as cost-efficiently and effectively as possible. Can you give us examples of how this consolidation would improve service delivery to Canadians?

[John Chang]

When you have the very federated structure that you have today, a symptom of this federated structure being relatively inefficient from a service delivery perspective, setting aside cost, is the presence of over 300 data centres and over a thousand points of presence, potentially. This means that underneath all of that, the service delivery is disjointed. It's not standardized.

As I mentioned earlier in regard to server and mainframe infrastructure and services being commoditized in the marketplace, I'm sure you've heard of service offerings like cloud. We very much see the evolution of these infrastructure services into a utility very similar to the hydro business in the 20th century. In the 20th century, some of the organizations used to have their own hydro departments. I think that over time you will see that these types of infrastructure services will become a ubiquitous utility. That means standardization, a high quality of service, and not having such a disjointed delivery model.

That's happening in the private sector. Through case studies and other examples, we see that in the public sector as well. That kind of standardization will enhance service delivery and quality of service, we believe, to your clients, who are the people of Canada.

[Ron Cannan, Conservative, Kelowna—Lake Country, BC]

I've toured a centre, and I know the--

[The Chair, NDP, Pat Martin]

Excuse me, Ron; you're well over time already. You'll have to wait for the next round, please.

[Ron Cannan, Conservative, Kelowna—Lake Country, BC]

Okay. Thank you very much. I appreciate it.

[The Chair, NDP, Pat Martin]

Next, for the Liberals, we have John McCallum.

John, you have five minutes.

[John McCallum, Liberal, Markham—Unionville, ON]

Thank you, Mr. Chair.

Thanks to both of you for being here with us today.

My first question has to do with the timing of the costs as compared with the timing of the savings. I notice that on page 21 you have the timeframe going out as far as the year 2026. I would have thought that the costs would generally precede the benefits. In order to make the changes, one needs to spend money in the earlier years.

Would it be true that in the earlier years there's a net cost and that at some point the benefits exceed the costs? If so, how many years out do you go before the benefits start to exceed the costs?

[John Chang]

Yes, you are absolutely correct that both in transition and in transformation you need to invest to reap the benefits. On our page 21, it's highlighted by the red bars you see.

In transformation, where you're actually consolidating the server environment, for example, you need to set up a virtualized server infrastructure to take advantage of that. That's an example. When you're in transition to actually move some of the work to internal shared services or to a private sector provider, you need to incur transition expenses to do that. We've documented those one-time costs there, first of all.

Second, in terms of the timeline here, this is for illustrative purposes only. As I just mentioned, we view this as a series of gates, and for each decision.... For example, virtualizing the servers across the Government of Canada would need to have its own business case and would need to have its payback that has been approved by the Treasury Board and other stakeholders. For transition and for the different service bundles as well, each one of those is a separate business case as to when the payback and the returns will occur.

You're absolutely right: in any transformational program like this, you need to make some investments.

[John McCallum, Liberal, Markham—Unionville, ON]

Based on what you know—I know you don't have the precise figures—can you tell us if it is on the order of two years, five years, or 10 years before there will be a net saving to the government?

[John Chang]

I think that's a decision the government has to make within the business case that it has built: what is the payback period of a specific bundle or specific subset of a program? That will be a decision point in the future. I can't really speculate on that.

[John McCallum, Liberal, Markham—Unionville, ON]

Is it conceivable that the payoff time could be as little as three years, or would it necessarily be longer than that?

[John Chang]

I think it would definitely range. Again, I'd be speculating, but in the private sector I can tell you that a three-year payback would not be acceptable to most people. I don't anticipate that in the government, perhaps; I'm not sure. My experience in the private sector is that on a server consolidation opportunity, for example, a three-year payback would not really be acceptable.

[John McCallum, Liberal, Markham—Unionville, ON]

It would not be acceptable.

[John Chang]

Yes, I think you need a faster payback than that. Based on the case studies we've done, and what we see organizations undertake, we anticipate that if it's implemented properly.... I mean, that's the critical part of all of this: it's all about execution. If you do it properly, we think those savings potentially can mirror what's happening with best-in-class organizations in the sector—

[John McCallum, Liberal, Markham—Unionville, ON]

You're making that statement even though other witnesses have told us that with all of these departments, this is hugely more complex than the typical private sector case.

[John Chang]

Organizations like General Motors or Hewlett-Packard are very large organizations as well. Some of our case studies have revealed that there are some best-in-class timelines available for returns. Again, there are going to be examples across the board in different case studies that can illustrate either end of that, but—

[John McCallum, Liberal, Markham—Unionville, ON]

Okay. I'm sorry to interrupt, but my time is quickly running out.

On page 90, you indicate savings in a range of $45 million to $293 million. That is a huge range. Can you explain why those savings are so imprecise or variable?

[John Chang]

I mentioned earlier that there are two sources and fundamental groupings of those savings. One is the transformational area and the other is the transitional.

In the transformational area, it's consolidation, virtualization, and rationalization of both facilities and servers, and potentially, as I mentioned earlier, the mainframes and storage and so forth. To do that analysis bottom-up—and I used servers as an example earlier—you need to understand what the capacity utilization of those servers is today.

There are over 25,000 servers in Canada. We didn't have time to actually do a point-in-time analysis of the capacity of those servers, so what we did, given our experience in other virtualization and consolidation engagements, was take an estimate of high and low, and that in essence provided the range.

It's the same with the transition. The actual benefits realized from that will be subject to implementation and the different timelines you have, so again, that's the range. I know it is a big range, but unfortunately that's the best we can up with.

As for our recommendation to you, as you go on this journey, these gates will come, and each one will have to have a very bulletproof business case, a detailed business case that includes, obviously, tighter estimates of what the actual savings are going to be, how soon they will be realized, and what kind of investment you need to make to have that happen.

As you saw at the end of our report, we actually make that point very, very clear. This is directional in nature, and you need to do a detailed analysis into the future of each of these streams or gates. That's just good, prudent management.

[The Chair, NDP, Pat Martin]

Thank you, Mr. Chang.

Thank you, John.

[John McCallum, Liberal, Markham—Unionville, ON]

Thank you, Mr. Chair.

[The Chair, NDP, Pat Martin]

For the Conservatives, we have Peter Braid. You have five minutes, Mr. Braid.

[Peter Braid, Conservative, Kitchener—Waterloo, ON]

Thank you, Mr. Chair.

Thank you, Mr. Milam and Mr. Chang, for being here this afternoon.

Mr. Chang, I wonder if we can start at a high level and really boil it down. According to your report, what are the top three benefits of the government's initiative for data centre consolidation?

I'm actually going to ask you a three-part question. First, what are the top three benefits? Second, what are the top three risks? Third, do the benefits outweigh the risks? Let's start there.

[John Chang]

One of the key benefits is to actually deliver on the requirements that we gathered when we engaged in January 2010. We gathered these through interviews and surveys with many stakeholders in the Government of Canada and we came up with 65 requirements. We've categorized them into 11 categories, which you see in that report. We did the options analysis against those, so what we're recommending best addresses those strategic requirements.

I'll read some of the strategic requirements for you just to give you the context. Within the cost and funding area, to limit capital-intensive expenditures is one of the things we were told is a requirement of the Government of Canada. To support predictable and sustainable funding models is also one of your requirements. Also, within availability, provide redundant data centres, as we were told that disaster recovery and business continuity was a requirement.

We have 65 of these requirements and we did the analysis against them. What we're recommending best fits these requirements, if implemented and executed properly. That's the first thing I would say.

Second, as part of the same exercise, we gathered 26 risks associated with data centre delivery today. We categorized them into six categories: financial performance, quality, flexibility, implementation, risk, and business alignment.

Again, we applied each one of those risks or categories against the options and came up with what we recommended: that the centralized model option has to be the least risky in regard to avoiding these risks that were identified during the data-gathering phase.

[Peter Braid, Conservative, Kitchener—Waterloo, ON]

Okay.

The third part of my question is with respect to your recommendations in the report and your final conclusion. Do the benefits outweigh the risks?

[John Chang]

Well, the benefits are going to be realized when it's executed successfully. Let me put it that way. Our report is directional, and it's on paper. Our five case studies and many other examples all speak to realizing benefits through a successful execution and not repeating some of the mistakes that others have provided before.