|
Page 3 of 5
Capacity planning
There are some tried and trusted strategies to manage data growth, the most obvious ones being Information Lifecycle Management (ILM), capacity planning, consolidation, and virtualization.
The value of information changes over time. When it is first created it is likely to have high value or critical status. Over time its value decreases, it is rarely accessed and moves to non-critical status. With an ILM policy, the critical data can reside on faster, albeit more expensive, Tier-1 storage such as fiber channel. Non-critical data can be moved to Tier-2 that uses cheaper SATA disks. Tiered storage and ILM policies are a way for companies to control storage costs.
A costly mistake that many businesses make is to purchase and deploy more storage without adequate planning. This storage may also be deployed in silos among the various departments in the organization, leading to a confusing tangle of heterogeneous storage systems and management software. Such environments would require IT staff expertise, at additional costs. Deploying storage in silos could also result in underutilized storage assets.
Srikant Chakrapani, Director, Enterprise Solutions, Hitachi Data Systems, opines, “Because of the way storage is packaged, businesses end up buying up to 75 percent more capacity than they actually need. It's cheaper to simply buy more storage than to hire somebody to manage it, so an organization's first response to a storage crisis will be to throw more storage at the problem.”
Parikh of F5 Networks has observed that the average utilization rate of file servers on NAS systems typically range from 30 – 55 percent. Many enterprises have a lot of unused capacity with no way to reallocate the capacity. “You may have multiple tier storage and in each tier you have utilization of just 30 – 50 percent. In another tier you may have 25 – 45 percent utilization.”
 Sanjay Lulla, Director, Technology solutions, India & SAARC says it’s important for customers to think through their buying decision and ensure that a sufficient number of spindles are deployed to meet performance needs. “Customers have been buying storage on capacity. Today 1 TB can be deployed with a single disk. But with storage you can get performance only when you stripe across multiple disks. As a thumb rule the best IOPS (Input/Output Operations Per Second) that you can get from a single disk is 125 IOPS. Earlier you’d use multiple disks to reach 1 TB. So today, the capacity is the same (1TB) but there is a huge difference in the IOPS (as compared to the collective IOPS of multiple disks).”
Amol Pathak, Global Head (Network Storage), Tech Mahindra, says most companies use just 25 percent of their current storage capacity, yet they continue to add capacity. “You must allocate/utilize the right kind of hardware as well as storage capacity for each application and database that you use. So instead of buying new hardware all the time, we can use the existing one itself if it is properly optimized.”
Some have also taken to consolidation and virtualization as a means of optimizing storage utilization, reducing the complexity and hence simplifying storage management. Companies are working towards consolidating their hardware infrastructure like servers, operating systems and storage.
Storage optimization technologies such as data deduplication, virtualization, and thin provisioning can also be used to check data inflation and control costs. Read the story ‘Storage technologies to beat the recession’.
Surajit Sen, Director - Marketing & Alliances, NetApp India, said, “Enterprises need to first make an assessment of their current asset utilization. Technology innovation in recent times, like thin provisioning, thin cloning, thin replication, advanced RAID, and deduplication have made it possible for drive up storage efficiency.”
Data protection and backup >>
l Page 1 l Page 2 l Page 4 l Page 5 l
|