What are the biggest concerns among CIOs for the data center in 2006? The Gartner Group in its 2006 survey of IT executives gives the following four as the top of the list.
1. Business Continuity & Disaster Recovery
2. Virtualization Directions & Technologies
3. Server & Storage Consolidation Trends
4. How to Run an Effective 24 x 7 Operation
To those priorities, Id add a fifth: data center command and control. But first, heres my take on where the industry stands on the four.
Business continuity and disaster recovery: There has been a lot of progress here. Spurred on by disaster (both manmade and natural), government compliance regulation and revenue streams increasingly tied to ecommerce systems being up and running, businesses have been investing a lot of time and money in this area.
Vendors are doing much better at selling systems that can not only recover from a total shutdown, but can allocate resources when individual system elements slow or fail.
For 2007, I think you will see more of this systemwide approach based on redundant arrays for inexpensive servers and storage distributed throughout the enterprise. System management and application development to take advantage of this infrastructure should be a top priority here.
Virtualization: Once it seemed that virtualization was only the province of operating systems. Now, server virtualization, storage virtualization and network virtualization are common topics among data center administrators.
The concept of virtualization is great. One of the biggest problems for data centers in the past has always been low system utilization and the high costs of simply maintaining those underutilized systems. Virtualization, where system resources can be efficiently spread throughout the enterprise, promises an end to the days of $10,000 servers operating at 20 percent capacity.
Software vendors are doing better at licensing their products to acknowledge this new virtual world. But there are caveats. The virtualization management tools can lock you into a vendor as tightly as the old days of proprietary operating systems if you are not careful.
Server and storage consolidation: Id think instead of consolidation, the term utilization would be a better description. Server utilization will have to increase (as I talked about regarding virtualization) if those dollars once tied up in system maintenance are going to be redeployed.
Storage is a more interesting issue here. Storage needs continue to increase at fantastic rates. And those rates will only increase again as video, audio, voice over IP and mobile devices become the common elements of enterprise computing. You will never consolidate your storage, but you might be able to bring reason to the ever-increasing needs through the use of virtualization, vast arrays of inexpensive disks (think Google), outsourced storage and tiered storage where archived data is sent to tape or storage systems that can be powered down while not needed. The vendors still have a lot of work to do.
How to run an effective 24/7 operation: This is a key management issue for data center administrators. You need to keep the system operating, replace your system administrators who are nearing retirement age and find an efficiency that wont drive power, bandwidth and personnel costs through the roof. The vendors should be playing a key role.
Rather than accept a scenario in which something breaks, you send out a beeper 911 to call in the repair troops and then start the long process of figuring out what busted, CIOs and data center managers need to put their dollars and resources toward predictive systems that can alert them to potential failures rather than sound the alarm only when something fails.
Redundant systems need to be inexpensive and fully capable of handling an increased load while failed system elements are replaced. And the system must be looked at as a total system, capable of taking hardware, operating systems, database systems and application systems into consideration.
Which leads me to my fifth element, command and control. While vendors have done a good job at adding control elements to their product line, few are looking at the value of bringing all products (including those of competitors) into the management infrastructure. Ive seen signs of those developments from stalwarts such as EMC, Hewlett-Packard and IBM, as well as from newcomers such as BladeLogic and Cassatt. This area of command and control will be, in my opinion, the hot topic for data centers in 2007 and a place where it would be well worth it to plan on applying your data center dollars and resources.
eWEEK magazine editor in chief Eric Lundquist can be reached at [email protected]