Sunday, December 12, 2010

Reducing Federal Data Centers by 38% in Four Years

Vivek Kundra, the US CIO, has just announced an ambitious program to reduce the number of Federal Data Centers from 2,094 by 800 in 2015, or by 38%. (see http://cio.gov/documents/25-Point-Implementation-Plan-to-Reform-Federal%20IT.pdf).

The number of data centers in the Federal Government is actually larger than 2,094, (see http://pstrassmann.blogspot.com/2010/12/how-many-data-centers-in-dod.html).

Though the Kundra is driven by pending budget cuts, the metric of counting how many data centers are eliminated can be misleading. Only dollars in the reduction of Operations and Maintenance costs are a proper measure of true reform.

Having presided during my career over data center consolidations at Kraft, Xerox and DoD involving hundreds of such attempts, I would be cautious about committing to 38% data center consolidations so fast.

Large data centers have the characteristics that they support a variety of customers. Each customer is connected to their servers with unique services provided in each case. One cannot just rip out a collection of servers and mainframes without the careful re-engineering of whatever links to it. These connections are mostly managed by sundry small contractors who provide maintenance and operating services, each using the best of custom-made technologies that they have available.

In any data center one will find multiple versions of operating systems, homegrown security appliances, unique communication processing methods, incompatible device interfaces and customized network management consoles. It is unlikely that all of the procedures in place are fully documented.

Resident contractors retain much of the know-how for operating the data centers, particularly under condition of failure. Potentially, the Government Data Centers can include up to 2,814 different server versions connected to 1,811 listed versions of clients that are managed by 1,210 versions of operating systems. If folded into any cloud, all of this diversity will have to be reduced drastically. How many of these are present in the existing Federal Data Centers is not known but my guess is that our legacy systems remain a large depository of software that has not been upgraded for years.

Every cloud contractor attempting to merge data centers into a vendor's Infrastructure-as-a-Service or Software-as-a-Service environment will have to first contend with the complexity they will have to be migrated out of an existing data center to a cloud vendor that always operates in a streamlined way. Each cloud vendor has defined a way of handling customers in tightly prescribed ways in order to preserve the greatest possible lock-up on the customer's business.

SUMMARY
Dictating the cuts in data centers without also extending to the consolidation to desktop hardware, desktop software and communication processing is risky.

Before proceeding with large-scale mergers that may reduce the number of data centers further consideration should be given to the disruption during migration to anywhere. Any plan to reform federal information technology data center operations should first invest into the engineering that would dictate standards how the new Federal data center environment will function.