Advertisment

Case Study: Server Proliferation

author-image
PCQ Bureau
New Update

Kasim Sayed was one of the old boys. From a hands-on Unix administrator he had become the CTO of a multi-crore organization. Although he liked to argue in favor of Unix on big iron, Sayed ran the companies heterogeneous network fairly efficiently, and in an emergency, he could actually roll up his sleeves and give his young network administrators a run for their money, on Unix or Windows.

Advertisment

What bothered him now was the proliferation of servers in the organization. At last count there were 37 machines across the country that the IS department classified as servers. These ran a variety of OSs including OS 400, Solaris, Linux, Win 2000, and there was even one machine with NT. These ran a variety of jobs, ranging from accounting, mail for the 1000 plus employees spread across 23 locations, (6 mail servers) an Intranet, sales applications and assorted other activities. Managing so many servers was becoming a big task and he did actually for the days when he could have telneted into a server over

adialup modem. Going by plans for the future, the number of servers could only increase.

His team was divided on what to do. Someone advocated consolidating all the servers into one location with wide pipes to all offices. Another argued that consolidation should be at the regional level. A third was in favor of letting the servers remain as such and instead install an enterprise-wide network management software. A fourth wanted to do away with the multiplicity of hardware and OSs and move to a uniform server spec and OS, but was not sure which one to choose.

Expert’s Answer

Advertisment

There is no definite “do this and all your problems will be solved” answer for this one. What we will do is look at the approach that Kasim has to take.

Server consolidation can have various approaches (see article in this issue on server consolidation page 20. It’s best if Kasim begins to identify these as well as their benefits and pitfalls.

Kasim’s first step in this endeavor would be to temporarily put on hold all current or ongoing server-related procurements and upgrades and to centralize the procurement and decision-making for all servers.

Advertisment
Fabian J D’mello 
Pre-sales Technical Specialist, IBM eServer xSeries

He should begin by conducting a comprehensive study of the current server setup. This study could be conducted in house by Kasim and his IT team or an external consultant could be appointed. Some server and storage vendors offer this as a free service in order to propose a solution, which entails their hardware and services. Such opportunities should be leveraged.

Once that is done, he should set down his objectives for server consolidation– the gains he plans to get from a consolidation exercise, and seek budgets for the same. Like with any other purchase (purchase is a major part of a consolidation exercise), he should consider various hardware, software and service offerings, evaluating their pros and cons to decide on his approach and should then begin with a pilot with a set time line and benchmark the results. 

Advertisment

He should implement a full consolidation solution if the pilot meets the objectives set.

Define objectives 



Given Kasim’s current scenario, his current biggest issue is management. However, during his study he would have noticed other issues like outdated hardware, orphaned applications that are no longer maintained by vendors, underutilized or duplicated hardware and so on. Kasim needs to identify all or a subset of these also as part of his server consolidation exercise. Having laid down these objectives, it’s now time for Kasim to garner top management support on budget allocations for the server consolidation.

The objectives can be outlined to clearly indicate current IT spending and the return on investment the proposed server consolidation solution would provide, within a measurable timeframe.

Advertisment

Implement a small pilot project



While the decision on a particular solution will have to be evaluated with multiple vendors, consultants and even peer references, it’s always best if experienced firsthand. Kasim should begin small by conducting a small pilot for a couple of departments or regions. This should be both time-bound and goal-based, with effective and realistic measurement against the objectives laid down at the beginning.

The result and lessons learned during this pilot would help highlight drawbacks and additional requirements if any. It would also help prove the ROI and with the final budget-allocation for the project.

Final implementation



If the pilot results meet the expected objectives, and issues and their resolution are all identified–then it’s time to go ahead and plan for a proper Server Consolidation in a phased and timed manner. Again it would help to benchmark at regular intervals to see that these objectives have been met. 

It always helps prove you were right, even if it was two years ago, especially to your Finance Officers!

Advertisment