A recent article in IndustryWeek discusses a revealing survey of enterprise datacenter decision makers. The survey found that while 45% of respondents have implemented virtualization solutions, very few are using automated management tools to improve efficiency and resource utilization. The problem is further complicated by lack of personnel expertise–meaning that all those “manual” operations may not even be getting done.
IT staffs already have their hands full with the number of hands-on tasks that must be performed. As new applications are implemented, they must be properly deployed, parameters and policies must be set, and the applications must be monitored for proper usage and performance. Adjustments must almost certainly be made until they are well in place and running smoothly. Elsewhere, daily fires must be put out where systems or applications perform poorly, incorrectly or not at all. When routine system management tasks are added into the mix, system administration can become a considerable burden.
Routine tasks such as backup, antivirus and anti-spam updates, defragmentation, system reporting and other such jobs should be as automated as possible. Aside from the fact that, not automated, they are consuming valuable IT time, there is the additional danger that such tasks become completely overlooked.
Some tasks can be scheduled to run automatically at certain times, and this is a partial solution. But the key word (and the truth of the matter) is “partial”; somebody must decide when these jobs should run and must actually schedule them. Again, are the tasks getting done, or are they getting done correctly? If it is possible to completely automate some or all of them, it is obviously vital that this be done so that already-overloaded IT personnel are not constantly distracted.
In one example, defragmentation, a task vital to insuring maximum system performance and reliability, is normally scheduled. But this job does require an experienced eye; systems and drives must be analyzed to ensure that defragmentation runs are scheduled on the right systems and with enough frequency to keep up with fragmentation. In addition to the question of the chore being done correctly, this is IT time that could be far better spent on more important work, such as new system and application deployment. Scheduling can be a hidden drain on time and resources.
Additionally, in today’s fast-paced computing environments, scheduled defragmentation has been discovered to be inadequate to the job; it is actually leaving fragmentation behind, and in some cases not defragmenting at all.
Defragmentation solutions are now coming on the market which are fully automatic, requiring no scheduling and which do not impact system resources. It is one task that can be completely automated, and it certainly should be. Research will help IT personnel discover other tasks which can also be fully automated, and a program to find and implement them would certainly be of great benefit to datacenters.