Tuesday, December 29, 2009

Hypervisor Star Wars

With the technobabble name of Hypervisor, I think that warrants a slightly science fiction post title, don't you?

Server virtualisation, which is achieved using one of the hypervisors on the market, is on the agenda of every CIO or CTO at the moment, whether that is for server consolidation or a grander move towards utility computing and the much vaunted cloud. So, I decided I would rub my twopenneth together and post where my thinking is at on this theme.

When I look at this sort of infrastructure technology I’m looking at the clarity of roadmap. Whatever decision a CIO/CTO makes, they’re going to live with it for probably about 5 years so when I’m reading about the big players (Microsoft Hyper-V, VMWare and Citrix XenServer), I’m not just trying to see who can do what now, but who is demonstrating thought leadership which will put them in good stead for the future.

Now, you don’t even need to go digging very far to analyse Microsoft Hyper-V… you just can’t rule it out. The biggest OS company in the world, and with many $billions sunk into datacentres that they need to start to build revenues off, they simply will not drop the ball on server virtualisation. You might as well value MS shares at mere cents if you do! Interestingly, they probably haven't the greatest feature set right now (they can't do real time migration), but they will architect their product really well so the features will arrive in due course and they will not be killed any time soon. Beyond that you can see how they will develop Hyper-V in the medium term. I can see an interesting triple play emerging with Hyper-V, Azure and all those datacentres.

That triple play brings us to VMWare. It took me far too long to get my head around why EMC bought them because I couldn’t see the synergy. Anyway, a month or two ago they announced a 3-way partnership between Cisco, VMWare and EMC called Acadia and all becomes clearer. This is really exciting. Not only can I buy virtualisation technology, but I can buy myself a private cloud. Aside from that (as if you need it), VMWare have good public domain information about their roadmaps and you can see how a client can go from legacy server infrastructure, through consolidation to private cloud and public cloud. They also have a very mature vision of what separation of logic and hardware can do for you and that sort of quality thinking will keep driving their product development.

Now with Citrix XenServer I see more uncertainty. With a Market Capitalisation of $7bn and profits of $150m, Citrix aren't, in my opinion, big enough or equipped with the right strategic partnerships to ride this wave to the next level. So, they will only exist if they are bought (in fact, I’d have thought that much of their MktCap is acquisition speculation) and until that happens, I wouldn’t make a 5 year investment decision on them. There are acquisition rumours, but it needs to be the right buyer. I don’t think Oracle would be the right buyer for me- I don't see any synergy as potent as that envisioned in Acadia for example. With all that, their tech may well be great but I’m not even going to look until there is that corporate stability.

A great book here is The Big Switch by Nicholas Carr. Treat yourself. It plots the future of utility computing out and takes it to places where really innovative companies are playing- like 3Tera (watch the demo on their website where you just build your infra in a Visio-style GUI). Yes, 3Tera is too bleeding edge for us in midsize corp for now, but it will pay to keep an eye on how their balance sheet develops.

A really exciting future no matter which mast you nail your colours to. The one true lesson we would all agree on is, don't do nothing.

Wednesday, December 09, 2009

Explaining service oriented architecture using the analogy of local economics

Modern enterprise systems are complex and differences in approach are, perhaps, well described using the analogy of a town’s economy. The traditional approach of buying functionality as packaged systems and configuring the last few features is like the communist approach to town planning where the vendor is the state. The state strictly defines the services in the town from the butcher and the cobbler to the grocer and the taxis. For the system to continue to work, almost all changes (except perhaps the colour of the baker’s walls) must be escalated to the state for review, evaluation and decision. When the state agrees a change is a good thing, it simultaneously makes the change across all towns in its jurisdiction. As the town grows, this approach increasingly restricts further growth until a state of bureaucratic paralysis and economic stagnation is induced- the fact that the town can be any colour is frustratingly irrelevant when all the towns are functionally the same and real differentiation is not possible.

In a system built to service oriented architecture concepts the state’s governance moves to the client side. In our town, the economy is liberalised and trusted to market forces such that individual services can be defined autonomously from state controls. Thus, an ecosystem develops. A service owner can choose to respond to market needs and is free to specialise, compliment someone else’s service or wrapper another service to add value. For example, a taxi driver could encapsulate a cobbler service to create a new “home delivery shoe repair service”. In theory the cobbler service can be modified and as long as the published “service definition” is up to date, the taxi driver can maintain the quality of their overall service without needing detailed information about how the cobbler’s internal operation works. There are few restrictions to even consuming services from other towns. The state still has a responsibility to deliver essential or generic services such as core infrastructure and a directory of the town’s services, but its role is mainly about maintaining the rules of the economy (=standards), rather than trying to deliver value added services. With limited state intervention, growth breeds further growth and the macro economy of the town responds organically to external influence, especially competition.

Tele-Wine Tasting

Cisco, BT and Chief Wine Officer combined some Christmas cheer with a technology demo wonderfully last night. The team used a telepresence room in Cisco's London HQ to host a wine tasting evening from similar kit in BT's New York office. This was a fantastic test of the technology due to the need for it to allow extremely nuanced communication between two groups of people. Only two further things are needed from Cisco to make the wine tasting as good as having everyone in the same room:

1. Object teleportation- removing the need for the wine to be sourced and stored at each location.
2. Time machine- it had to be remembered that it was just 1pm for our valient hosts, who matched us glass for glass (except for the sommelier who diligently spat after each sip).

So, nearly there Cisco. Piece of cake.

Meanwhile, until such tweaks are made, a telepresence room has been set up by Cisco at the Copenhagen Climate Conference (and will stay for about a year thereafter), both to facilitate collaboration and also as a great demonstrator of a technology with the potential to reduce the need for travel.