EMC -- with its Virtual Computing Environment (VCE) Vblock -- is driving the integrated IT stack concept as hard as it can. But that's simultaneously helping Dell, HP, IBM and Oracle build stronger marketing positions against it.
Let's not argue the case for or against integrated IT stacks but instead accept them as a good thing. The idea is that enterprises and service providers increasingly have to buy masses of IT resources -- servers, storage and networking -- to satisfy cloud service demands.
They need to be able to repeatedly buy scalable IT platforms to respond quickly and easily to increases in service demand. It is simpler for them to acquire and operate integrated sets of servers, storage and networking than to have to buy the bits separately and make them all work together in their data centre.
The position is complicated by server virtualisation, which means having another layer in the stack to worry about. Having a virtualisation or abstraction layer in front of servers and storage and networking hardware means you can reallocate capacity to suit the application load.
EMC saw the need for this, as did Cisco when it introduced its UCS servers. That is how the VCE -- or VMware, Cisco and EMC -- alliance came into being with its Vblocks, integrated Cisco servers running VMware, Cisco networking and EMC storage. MTI has just sold the first Vblock installation in the UK, and everything in the VCE garden is looking rosy.
However, as is always the case with roses, there are some thorns.
It's obvious that the more integrated the stack is, the better. If a single supplier controls technology throughout the stack, individual component development to enhance coordinated stack processes can proceed more quickly.
Look at HP. If BladeSystem or ProLiant servers need a particular networking technology to best address storage, HP's ProCurve networking and StorageWorks storage products can develop it and test it quickly because it's all in-house. HP in-house management software can monitor and manage it, while HP diagnostics software can track performance and raise error alerts to the management software.
All of these interconnected things can be understood, defined, developed, tested and implemented far more quickly if the whole thing is owned, controlled and put together in your own integrated IT stack "factory."
Oracle is in a roughly similar, if not better, position compared to HP in that it owns application software, middleware, the Solaris operating system, Sun servers, Sun networking and Sun storage. It even has its own server virtualisation stack. All in all, this is potentially a formidable offering that could develop into the most integrated IT stack of all the vendors, putting HP in second place.
IBM can also optimise its servers and storage readily, but is less able to optimise its networking products as these come from third parties. It can provide integrated stacks with its AIX operating system, too.
Dell, with its OEMed technology, is in a worse position than IBM as some of its storage is OEMed from EMC and has to work with EMC.
We'll ignore the position of vendors that have only a single layer in the stack, such as NetApp and Microsoft. They are pushing an integrated stack message but necessarily through alliances. Indeed, there is a VCN marketing alliance (VMware, Cisco and NetApp) with its own templates for integrated IT platforms. However, NetApp is unlikely to build its own complete stack offering by buying server and networking hardware technology. The same goes for Microsoft, which is even more unlikely to buy server, storage and networking hardware technology.
The hardware basics matter here. The server vendors that own their own storage can integrate servers and storage. In principle, they can do this faster and better than EMC, which has to work with its alliance partner Cisco, which in turn will be predisposed to support open interfaces to maximize UCS server sales potential. EMC can't simply tell Cisco to add features to the UCS product. At HP, David Donatelli runs servers, storage and networking, and can tell people to get things done.
EMC is actually in quite a poor integrated stack position, being just one vendor in an alliance. If Oracle and HP execute well on their greater integrated IT stack potential, they will make more sales than Cisco, Dell, EMC and IBM. The only way these four trailing vendors could catch up would be by getting hold of their missing integrated stack layers and bringing them in-house.
For EMC that logic would impel it to have a much closer relationship with Cisco or to get its own server product line by -- the obvious back-of-the-envelope choice -- buying or merging with Dell.
I repeat: If integrated IT stacks are better for customers, the more integrated layers there are in the stack, the better. Again, the more layers there are that can be integrated by a single vendor, the better. The logic is undeniable.
The large and successful vehicle companies, BMW, Ford, Mercedes, etc., all have their own engine, gearbox, suspension and body shops. They own the main layers, as it were, in the vehicle stack and would be weaker if they did not. Surely the same industrial logic will apply to the production of integrated IT stacks, and, by that measure, Oracle and HP are leading the integrated stack race, with Cisco, Dell, EMC and IBM following some distance behind.
I'm sure HP and Oracle appreciate this argument and are both pumping it out in marketing messages to their potential customers.
All the marketing messages put out by Cisco, Dell, EMC and IBM in favour of integrated stacks are also generally strengthening the market position of HP and Oracle. Therefore, EMC in particular, by leading the integrated IT stack marketing drive, is helping to make its main two competitors stronger and weakening its own position. What an odd world we live in.
Chris Mellor is storage editor with The Register.