Prior to conducting an audit, most data center managers believe that their storage utilization is around 70%. Those who do rigorous audits (and there aren't many of them) are likely to find that their storage utilization is closer to 10%. My own experience interacting with data center managers at enterprises large and small tells me that 30-40% percent storage utilization is actually on the high end. That's not to say that these companies are only using 10% of their storage arrays -- quite the contrary. Their storage arrays are 80% full. However, much of that data is inactive.
Clearly there's significant room for improvement -- and cost savings. Before you commit precious budget resources to buying more storage hardware, be sure you're getting the most possible value out of what's already deployed in your data center. Evaluation and management are key. I recommend these five steps:
Five steps to maximizing disk capacity
1. Determine how much storage capacity you have.
You may choose to undertake a comprehensive evaluation of your data center, or you may start on one site with one database, or anywhere in-between. Once you've determined the scope of your audit, you can begin. You need to know what hardware is out there, including hardware across the Storage Area Network and hardware that's directly connected. You will likely be pleasantly surprised at how much storage you actually have.
2. Assess how much storage you're currently using.
Perhaps you have 75 TB and you're actually using 30 TB. You'll need summary reports of utilization, as well as the detailed drilled down views in order to identify over-provisioned storage, recover wasted storage and increase overall storage utilization. If you have the right data on historical storage utilization, strong and accurate trending reports and the right tools to do advanced mathematical projections, you can create precise capacity trending reports to give you an idea of future requirements.
3. Evaluate different types of data.
You need to learn whether you have old files, unimportant files and mission-critical data all sitting jumbled together on the same hardware (which is usually the case). Perhaps one area of your data center is the culprit in this inefficiency, or perhaps it is a pattern that runs throughout your organization. If you can isolate the problem, you can trace back to the source and solve it.
4. Devise a plan for tiered storage.
After completing steps one through three, you'll be armed with the information that will help you set policies. You'll be able to decide which types of files are critical and therefore need to be on the most accessible, best-managed storage, and which types of files can be moved, archived or purged. Many companies develop tiers of storage, segmenting different levels of data based on type and content. This is a critical step along the road to more efficiently and effectively maximizing your storage capacity today. The key to success here is to leverage what you already have. Find a vendor that can set up a tiered storage strategy for you without requiring new hardware purchases.
5. Implement your policies.
Several tools exist that can quickly identify and move files between your various tiers of storage based on the importance of each file. This general concept isn't new, but there are new technologies that may be useful for this step. Again, look for technologies that let you leverage the storage you already have. I have found that the best file location policy tools are those that reside in the file system, where the most detailed information about a file is located. File system policies are also transparent to the user or application and provide the best performance.
As you implement your policies, you will begin to group critical data onto the most accessible and best managed storage, and archive or delete unneeded files.
Shrewdly using the right tools to drive a strong return on investment will give you the breathing room to take those IT dollars and put them where you truly need them. And that makes everybody happy!
For more information:
Tip: A quick and dirty way to increase a RAID-5 array
Products of the Year 2004: Disk and disk subsystems
Tip: A better way to predict disk failures
About the author: Robert Soderbery is vice president, product management at Veritas Software Inc.