Hu Yoshida

Shirt Sizes are not the Way to Buy IT Infrastructure

Blog Post created by Hu Yoshida Employee on Nov 11, 2014

Jul 17, 2014


Last week we had an internal meeting in Santa Clara with representatives from all the geographies in Hitachi Data Systems. Since last week was also the finals for the World Cup in Football, we all received football shirts emblazoned with the words “One Hitachi”. As you can imagine, with an international audience, one size does not fit all. Prior to the meeting we all had to submit our shirt sizes, S, M, L, or XL. But even then, a size M in the US or UK is not the same as a size M in Japan or Hong Kong. Luckily we had enough of an assortment of sizes that we could swap around to find the right fit, and the material had enough stretch to accommodate some of us who did not have the proportions of a football player. The problem with shirt sizes is if you choose the wrong size (or if you outgrow it) you have to replace it with a new one that fits.


Also during the worldwide excitement over the World Cup, you may not have noticed that EMC announced their new flagship storage array, VMAX3, which they plan to deliver some time later this year. There were the expected additions of processor cores, storage ports, and storage capacity with faster feeds and speeds. What surprised me was that the VMAX3, now comes in shirt sizes, 100K, 200K, 400K, or small, medium and large. The 100K uses the Intel Xeon ES 2620-v2 processor with a 2.1 GHz 6 core processor, the 200K uses the Intel Xeon 2650-v2 with 2.6 GHz 8 cores, and the 400K uses the Intel Xeon 2697-v2 with 2.7 GHz 12 cores. From the material that was published at their announcement, it appears that the upgrade from one size to the next is what we used to refer to in the old days as a “forklift” upgrade. Within a shirt size like the 100K, you can scale up by adding another “engine”- up to four with the 200K and up to 8 with the 400K. But these engines have a fixed amount of ports and cache per engine that you get whether you need them or not.

This is in contrast to our VSP G1000, which can scale from small to extra large by non-disruptive addition of Virtual Storage Directors, front end and back end directors, switch and cache modules. The only requirement is that these components are added in pairs to ensure redundancy.

The prior week, we hosted the new CIO of a large retailer at our Executive Briefing Center. Being new to his job he was touring Silicon Valley to get an update on the latest technologies. He was particularly interested in our Unified Compute Platform. His predecessor had purchased a VCE. VCE is also sold in shirt sizes so the former CIO opted for a small shirt size since it was preconfigured and easy to install- about 3 months. Unfortunately after 3 months, they found they needed a larger shirt size and were looking to buy another converged compute platform. That would not have been a problem with UCP since it can grow from small to extra large and beyond. If you underestimated your infrastructure requirements – no problem. With UCP we can preconfigure a system to facilitate the initial install, 3 to 5 days, and we can expand it non-disruptively as workloads and business requirements change.

Buying infrastructure is not like buying shirts. It is hard to determine what is needed even a few months from now, much less 5 years from now. The infrastructure you invest in today must be able to scale to meet your business needs. You also want to be able to right-size it so that you are not overbuying. Hitachi’s approach to infrastructure enables you to do that.