It's a simple concept, scale down operating expenses in your data center through consolidation. The strategy usually entails migrating older hardware and software technologies into new gear, or improving management strategy around existing systems architectures.
The benefits of consolidation are fairly well established. A recent report from Boston-based AMR Research Inc. concluded that taking this approach typically delivers savings between 20-30%.
However, when you push beyond the obvious economies of scale, it's clear that data center consolidation remains a tricky process. Mike Chuba, analyst at Stamford, Conn.-based Gartner points out that using consolidation as a means of managing total cost of ownership (TCO) can often demand significant upfront spending.
"A lot of people are blind to the fact that sometimes to achieve savings you have to first spend money," he said. "Upfront costs often derail consolidation projects when managers discover that they can't get financing for the necessary hardware if they haven't pre-quantified what the long term savings may be."
Based on the reality that consolidation often doesn't come cheap, every business needs to approach its own data center with a critical eye toward whether consolidation truly offers measurable benefits, and how the process might work best in meeting individual needs, said Chuba.
The AMR report, "Data center consolidation: Opportunity in operational excellence," stresses
Planning is everything
Every consolidation project is fundamentally different based on goals within an organization and the structure of the data center itself. In order to prove the potential value and impact of a project up front, it's necessary to sketch out a detailed schematic that considers many different possibilities of how to get where you want to go.
Cindy Borovick, analyst at Framingham, Mass.-based International Data Corp. recommends that businesses approach data center consolidation on a case-by-case basis.
"You want to find a case study of someone doing consolidation, as well as find a case study of someone building out infrastructure, just to weigh all the options," she said. "The secret is finding what's right for a specific business in the long term. There's a need to balance short-term savings with long-term investment sensibility."
Borovick said companies must consider a raft of issues related to potential consolidation including availability of products and the ability to have real impact on operating expenses.
The devil is in the details
Gartner's Chuba is emphatic when it comes to reinforcing the idea that consolidation doesn't usually come easily.
"Users often say they've been mandated to cut costs 20% and think it's simply a matter of downsizing by one-fifth, and that's not necessarily the case," he said. "Consolidation doesn't magically happen, it's not moving boxes out of a room, the only way it happens is if someone gets their fingernails dirty."
Chuba points out that a project manager needs to walk through every existing data center software contract an organization has signed, to figure out if the agreements allow for true savings via consolidation.
"You have to determine whether each contract is positive, related to issues such as scaling toward a per-processor paying strategy," said Chuba. "It's not always easy to figure out if consolidating to one, per-processor price is doable, or how much money can truly be saved within the guidelines of existing agreements."
Grease the wheels
According to AMR, when cost savings are the primary driver of consolidation, rather than improvement of data center continuity, projects are often thwarted by greater internal scrutiny. And not just from the top down. Even when an initiative garners corporate leadership's stamp of approval, resistance may crop up on the business unit management level. These managers often view consolidation as a loss of local control over service and spending, which it likely may be.
For this reason AMR recommends selecting a project manager that demonstrates the political skills necessary to marshal sufficient organizational support. This can mean anything from appointing a trusted company veteran who knows the affected individuals in the company, to bringing in an outsider with no emotional attachment. Consolidation often results in lowered data center headcount and the art of downsizing is a well-established balancing act.
Pick low-hanging fruit
Chuba endorses the classic approach of tackling easier consolidation issues before moving onto larger ones. Not only do small victories get the process moving in the right direction, he said, but they also help provide an indicator of the resistance an organization might evidence toward larger initiatives later on.
For example, Chuba said when planning consolidation many users "believe the fallacy" of one server per application, when there may indeed be 4-5 servers supporting an individual system. And it may not be easy to realign this architecture, he said.
By determining which applications can afford to move to single server architecture first, it becomes easier to foresee which areas of the data center might provide the greatest challenges, or cannot be shifted at all, he said.
"Any single application may have a testing, development or disaster recovery demand that dictates how consolidation should be approached," said Chuba. "A lot of the time this has to do with how disciplined the organization was in terms of its original procurement."
Chuba said in many cases with applications that are operated by a specific business unit the procurement process may not have been as not as disciplined as with the rest of an organization's data center. Gathering details on where the inefficiencies lie throughout the data center will point to areas ripe for consolidation, he said.
This was first published in April 2003