Data storage — Faster and Cheaper or Something else?

Andy Mulholland, CTO, Capgemini

Sometimes I find myself in a strange situation. I have seen a number of announcements and feel that they are important but I can't quite tie them together in my mind as a coherent view. I feel like that right now looking at the powerful product announcements from major vendors that compel me to believe I am seeing an industry on the move towards a real transformation around delivering capabilities in a new way. Okay call it cloud computing if we must, but that's too abused a term to really describe things.

Right now for the second time I am carefully reading through EMC's biggest ever announcement of product releases; a whopping 41 new individual products in one go. My favourite source for understanding EMC is Chucksblog by Chuck Hollis, EMC's Global Marketing CTO and sure enough he has posted a suitably massive 13 page missive about the 'largest ever product release in storage industry . It's a well laid out explanation that covers how even during the last couple of years storage growth kept right on rising, and how EMC has broken through various barriers to be able to get new performance points in every category. If you are in the market for more storage it's well worth reading to get perspective on the challenges and how EMC is rising to it.

It wasn't until I read an interesting document from the Enterprise Strategy Group on the future of storage in a virtualised data centre that I realised the question nagging at me was how we design storage into our systems, and not into our data centres? The point that this document makes clear is that our approach has been to support mainframes, and latterly servers, using data around applications, meaning the connection format has mostly been based on fibre channel. Conversely the servers that are being virtualised to form the basis for the data centre of today and beyond are based on Ethernet.

Suddenly the importance of what seemed an unimportant press announcement in the last couple of weeks from Intel seems much more important. Intel is providing free software that allows fibre channel to be carried over Ethernet to allow existing storage systems to gain a new lease of life supporting virtualised server environments. There is a nice video to explain aggregating server-based Ethernet and fibre channel-based storage networks on the Intel site. If you go there you will find that Intel has a really very good video library on many topics. Try the one on reinventing the computing experience for a look ahead as well as efficiency in the entire data centre .

The conclusion of the Enterprise Strategy Group report is that the immediate mismatch is time. Time to set up a virtual server and manage it is several orders of magnitude better than current benchmarks for setting up and managing storage systems. It's a good point, as is the Intel contribution about design and connection, and with that in mind if I now go back and read Chucksblog about why EMC believes its new VNXe products is a breakthrough to a new approach to server storage systems I think I get it. It might be me, but I often struggle to relate feature sets contained within product releases to the issues. More accurately I should say to the new issues that we are facing in how we build, operate and deliver resources in new ways to support new business demands.

Given the investment in an existing data centre we simply have to figure out some of these issues and get reuse out of what we have. There is a tempting alternative and certainly a quick fix for short term immediate business requirements in one of the other big product announcements. The partnership between Microsoft and HP delivered its first results in the form of 'Four Converged Applications to Enhance Business Results'. That means some very nicely packaged appliance style solutions that offer a fully virtualised set of HP computing resources covering the server and storage side, pre loaded and optimised with Microsoft software based around SQL Server etc. Amazingly I can't find any in-depth post on this apparently important move, just the press announcement , but at least in context at the Database Trends and Applications website.

The key feature, or proposition, is to deliver time, simplicity and reliability in business intelligence and real time data analysis, two of the most sought after new business requirements. The press release claims it 'can be set up and made operational in less than one hour'. So there you have the options; upgrade the data centre as part of a strategy to shift towards supporting business in a new way, or deliver a specific business requirement with its own budget. Either way the issues at stake are more than a comparison of product features!