The best part of my job as CTO of Hitachi Data Systems, is to meet with customers, developers, vendors, regulators, and analysts to share ideas and try to understand what is and will be required to store, secure, and access data. To gather this data I have become the unofficial Chief Traveling Officer for HDS. In order to be more productive in gathering this information and communicating with a larger community, I have decided to launch this blog to share my observations in the hopes of getting your feedback. Any thoughts on storage, data, content, and yes, information as it relates storage, data and content is welcomed and appreciated.

To start this conversation I  begin with the following observation.

At one time the great minds argued about how many angels could dance upon the head of a pin. A similar question today is how many terabytes can be managed by one full time equivalent. Toward the end of the 20th century, the storage pundits, were offering numbers like 5 to 10 TB per FTE. Others were putting this in terms of dollars with guidelines like $3 to $5 for every $1 spent on storage hardware. This stuck in most of our minds, that managements costs would be increasingly more costly than hardware, since storage costs were declining about 30 to 35% per year and storage capacity was growing at 30 to 50% compounded. More capacity meant more management, and more management meant more people, and the cost of people was expected to increase. These numbers were used to justify the need for investments in SAN, in larger monolithic storage arrays, and more intelligent software to offset the cost of management.

Recently I met a CIO who showed me numbers that contradicted this line of thought. His numbers showed that his storage hardware costs were 2/3 of his total cost of storage! This completely contradicted the conventional wisdom that hardware was only 1/3 to 1/5 the total cost. Was he one of the few successful CIOs that had gotten his storage costs under control, through the use of SANs, large monolithic arrays, and intelligent software?  Unfortunately,while he had done all these things, his growth in storage hardware was becoming irrational. It did not track to his business revenue. he was at the knee of that compounded growth curve where the rate of growth was headed straight up.  Rather than having his storage growth under control, he believed that his storage growth had become irrational. He summed up his storage problem in four sentences. 1) He was buying too much storage with growth rates over 100%. 2) He was paying too much with over 70% of his storage in high priced monolithic storage. 3) His utilization of storage was too low at about 20%. 4) And because of his irrational growth he was running out of power, cooling, and floor space in his data Center.

Was this an anomaly or were there others CIOs facing the same predicament? Since then I have talked to a number of CIOs and IT directors who have told me that they are managing their storage with the same people they had four years ago. If they had 100 TB four years ago and they had a 50% CGR, they now have over 500TB today and are attempting to manage it with the same number of people. Their budgets have been flat and it was too costly to hire and train new people. It was more cost efficient to just buy more storage. Some large accounts say they are managing a peta byte with 4 people. Many are growing capacity so fast that they have a standing order with their storage vendor to buy their latest and largest storage system every quarter. This a sure way to fall into the irrational growth spiral.

How does one rationalize the growth of storage? The first thing that needs to be done is to understand the data requirements with a comprehensive storage resource management tool. One of the hottest selling products recently for HDS has been the HiCommand Storage Services Manager product. Many companies are turning to this tool to determine their real storage requirements. The second requirement is a storage platform that can virtualize the storage platforms and enable dynamic migration for consolidation and technology refresh, to  increase the utilization of storage and reduce the requirements for power, cooling and floor space. The third is the ability to attach lower cost tier 2 and 3 storage behind a Tier 1 control unit to eliminate costly monolithic storage capacity costs and still have the benefit of tier 1 functionality.

What are your thoughts on the growth of storage versus the ability to manage it? Have we become so efficient that we can manage 200 to 300 TB per FTE? Are we sitting at the knee of the growth curve where growth becomes irrational? What tools do we need to manage compounded growth rates of 30%, 50%, 100% and more? Are there other CIOs who are in the same situation of irrational storage growth?

Comments RSS