India User Conferences and Operational Costs
by Hu Yoshida on Feb 7, 2009
This past week I was in India speaking at user conferences in Bangalore and Mumbai and meeting with our customers and partners. Our country manager Natarajan Viswanathan has been doing a fantastic job in growing our business there.
This was the first time for me to visit Electronic City in Bangalore. There I was able to meet with Mr. Shivshankar, vice president of the Infosys Information Systems group and tour their state of the art campus. I was also given a tour of their Network Operations Center which supports our IT help desk.
The topics that are of the most interest there is the cost of storage and disaster recovery in view of recent natural disasters and the terrorist attacks of 26 November in Mumbai. I had the pleasure of meeting Mr. Anil Jaggia, the CIO of HDFC bank who happened to be an eye witness to both the events of 9/11 in New York and 26/11 in Mumbai. You can be sure that HDFC is clearly committed to security and disaster recovery.
While the economy is a concern here, the impact does not seem to be as severe as in the rest of the world due to the conservatism of their financial system and the consumer demands of a large emerging middle class. There is still a very keen interest in reducing storage costs and increasing the efficiency of IT operations.
One of the misconceptions in countries like India and China is that operational costs are lower due to the lower labor costs of FTEs (Full Time Employees). This is no longer true. Today operational costs are more about the cost of time than the cost of people.
In the past, there was a lot more dependence on people costs in running operations – mounting tapes and disks packs, bursting jobs off the printer, and managing run sheets for batch jobs. Prior to SAN’s when storage was direct attached, you could spread the job of provisioning and managing storage to each server. Today, these tasks have been automated or networked in such a way that adding more people to an operational task does not help to complete it any sooner. The cost of migrating TBs of data from one storage frame to another is gated by the time to transfer each byte; the time, when and for how long, the applications can be suspended for that transfer; and the timing around warranty, licensing, and lease expiration costs. Adding more people to do the migration will not significantly reduce the time or the cost of data migration.
In fact operational costs in some verticals may be higher in places like India and China. Having had banking experience in New York and Mumbai, Mr. Jaggia explained to me that the difference between banking transactions in India and in America is in the value and cost of a transaction. For instance the average deposit in a rural area of India may be $100, while in the US it may be $1000. The same operational procedures are required to serve the $100 deposit as it takes to serve the $1000 deposit. Therefore the operations cost to service a million dollars of total deposits may be more than 10 times the cost in the US while the hardware costs for storage may be the same.
Globally, the focus of storage economics needs to be on the reduction of operational costs and the technologies to address it like virtualization and dynamic provisioning
Your observations are correct. The cost of acquisition, or labor, or time to provision are all dependant on the region, business area, value of data etc. It is impossible to generalize on the TCO elements, which is higher, lower, or even which ones are relevant on a global scale. That is why every time I work with an IT planner or architect we have to sit down and look at their total costs (from the baseline of 33 options) and then prioritize and provide quantitative measurements around those costs. Only then can we map the benefits of virtualization, dynamic tiering, de-dupe etc. to help them plan on reducing THEIR total costs.
Your comments about time vs. labor are also compelling. In the area of provisioning, there is less interest in the hours required to provision capacity compared with the wall-clock time the busienss or applications group have to wait for the capacity to be presented to the hosts. For many customers, this wait time can be measured in weeks, even months. With advanced, dynamic and flexible storage architectures (with or without a utility framework), this can be measured in hours. You may be aware of a european client that have put the provisioning task to an automated solution where the apps provision themselves. Now the time to wait is directly impacted, and that part of the TCO approaches zero.