The Storage Economist

A year or so ago, my colleague Claus Mikkelsen made a bold statement to storage architects that in their planning they had to start assuming the price of disk to be at zero. Certainly at HDS, eyebrows were raised, but he is right, the price of the drive itself keeps dropping. When measured in the old-fashioned-way of $/MB the price is below a penny (I found this old graphic, and by now we should be ‘underwater’ relative to price-per-MB).

Whether Claus made this comment or not, it is now HDS urban legend and I credit/quote him for stating it years ago.

I was in a global summit meeting this past week, strategizing on extending HDS’ economically superior storage architecture theme with our cloud-enabling and cloud-producing storage products. As we discussed TCO comparative methods the notion came up to “just ignore the cost or price of the disk drive.” Everyone in the room sat silent. We are a storage company; how can we ignore the price/cost of the drives? Heresy? Insanity?

This is where Claus’ comments came into my mind. If storage architects/planners assume the price to be approaching zero, why not do the same with economic models? For so many years the count and amount of disk drives (5.25, 3.5, 2.5 inch; Flash disks) determined the price. It seems to me that the drive itself is not a major differentiator when it comes to TCA or TCO. Many vendors put the exact same drive into their enterprise storage arrays anyway.

So what if we did entirely ignore the price of the drive, what would be left? Well in terms of acquisition costs, there are a lot of other ‘things to buy:’

• Cabinet for the array, power systems
• Controllers (nodes)
• Cache memory
• Front end and back-end ports (internal and external)
• Specialized software for data management, data movement, de-duplication, virtualization, thin provisioning, replication, copies, array management

And let’s not forget the ongoing and recurring costs- the financial gifts the keep on giving:

• Monthly/annual hardware and software maintenance
• All the costs of labor to manage, configure, tune, troubleshoot, backup, protect and decommission storage
• Power, cooling and floor space
• The oft-forgotten cost of migration
• Backup, tape, DR, long distance circuits
• Local SAN and data network infrastructure
• And the list keeps going on (click here for the complete list)

We know there will be price and cost differences with the more-expensive flash drives, enterprise-class FC SAS and large capacity SATA, etc. But let’s think beyond the obvious. Quite often, especially with new scale-out and JBOD/node architectures, the costs of interconnect, memory and the nodes (or controllers) will eclipse the price of disk in a TCA. Further review of the total architecture and solution is essential to make fair, balanced econometric comparisons.

Tweet

Posted in Cloud, Storage Best Practices, Storage Economics, TCO, Virtualization

Comments (3 )

Post Comment

David:

Excellent insight. This argument reminds me of a similar argument we had when high bandwidth data networks were becoming available for consumer market. The argument was that the price of voice will be zero as more service providers included voice as part of their data networks. The service providers were urged to move to provide more value added services such as video, teleconferencing, broadcast TV etc. Today, voice indeed is free but we pay a lot more for our communications than we used to in the 1980s to access social networking services, HD TV etc. Service providers who did not move from Plain Old Telephone Service (POTS) to Pretty Amazing New Services (PANS) predicted by Negroponti of Media Labs, have learnt a very expensive lesson by obsoleting themselves.

Storage vendors are currently in a similar situation. As server technologies are radically transforming the way the next generation data center is going to be virtualized, storage vendors are caught in a similar dilemma that faced Frame Relay vendors when SONET was introduced and ATM switch vendors faced when IP networking became a viable data networking solution with voice thrown in for free. Fibre channel SAN solutions, point to point replication services and many of the benefits that were once essential become irrelevant in the new virtualized data center architecture where virtual applications are decoupled from the hardware on which they reside. Servers, and storage devices will be like commodity infrastructure just as washers and dryers are in a Laundromat. Virtual applications fly-in and fly out, turned on or turned off based on workload demands and business priorities. Work-flows will be implemented with distributed business intelligence and the services will have no lotyalty to physical infrastructure. Physical infra-structure will be like buildings for rent.

To realize this, only one piece is missing – Commodity storage hardware with on-demand (real-time) Fault, Configuration, Accounting, Performance and Security provisioning. Once this is available, all storage management systems will migrate to virtual cloud where resource management is uniform across server, network and storage resources. This will put an end to current silo management systems and added complexity. Current hardware assisted virtualization in multi-core, multi-CPU servers along with Microsoft Virtual Hard Disks are pretty close but not quite. The data center refresh is inevitable given the performance of new server infrastructure. Question is who will win the storage war?

I have been thinking about similar issues and discussed them in my blogs
http://mikkilineni.wordpress.com

I would very much love to see your comments. I sincerely miss working with you

Best regards

Rao

[...] couple of days ago I was cited by one of my “buddy-bloggers in crime”, David Merrill (yes, he asked me first) for an outrageous comment I had made a few years ago, and continue to [...]

[...] 3.  Try to Ignore the Cost of the Disk Drive [...]

Post a Comment