HP’s community blog has a post on what they think will be hot in 2009.
- Power: We all got stung by the increased power costs in 2008. As usual, when oil went up, power costs immediately went up. When oil costs came down, power costs stayed up. Unfortunately, that’s the trend for the ongoing future. We need to get more out of every watt we consume. In Ireland we saw a little bit of marketing efforts about IT power consumption last year. I think power consumption will become a decision making factor in 2009. I know in 2008 we were doing that, we went with one model of Cisco switch over another because of it. We went with blade servers because of it. And we deployed machine virtualisation because of it.
- TCO: "Total Cost of Ownership" is one of those annoying consulting phrases from the mid-nineties. It’s back. Things like cloud computing and software-as-a-service have brought TCO back into play. We’re re-thinking about the need to own and manage a server(s) because we need an accounting system. Why build a power station in your back yard when you just want to turn on a light in the kitchen?
- Capacity Planning: Money is tight and those people who are deploying systems want to deploy them right, first time. They don’t want a machine that’s over-spec’ed or a solution that needs to be re-engineered because it’s not up to scratch. Microsoft has made some efforts with System Center Capacity Planner. I recently mentioned the volume sizing tool for DPM 2007. You should have a look at my posts for Hyper-V on RAM sizing and LUN sizing too. Beware of dealing with service providers where you only get to talk to salesmen. There’s absolutely zero capacity planning going on when you deal with them.
- Packaged Infrastructure: I think HP are selling things like pre-built blades/storage/clusters here. I’ll agree with it. Buying in a pre-built solution is much easier than buying servers from A, racks from B and storage from C, only to find incompatibilities, missing components and engineers saying "they’re responsible for that, not us". Again, you might want to look at managed server hosting where all of this is taken care of for you.
- Unified is hot: Welcome back to the mid-nineties 🙂 Back then we used to hear about things like CA Unicenter as being the unified solution for managing an IT infrastructure. I consulted on that stuff and it was good, as long as you loved patching … every week and on every machine starting directly after installing the software to get core functionality working *exhausting*. People got burned and along came open source. Then the point solution was the buzz. We’re back again at unified solutions. We have HP servers, SAN fabric and storage and it definitely works out better than a jumble. The integration makes us flexible …. fungible even (thanks Dave for that word!). We have integrated management thanks to Microsoft System Center, e.g. health and performance from OpsMgr and virtualisation from VMM. That’s all tied together using Active Directory. For h/w management I have one interface. For everything else, I have System Center.
- Performance per sq ft, per dollar per watt is HOT. Moore’s Law is NOT: Agreed. For 90% of servers, a single 4 core CPU is more than enough. We’re all about getting more from less now. For load balanced servers we can get more out of running fewer machines with x64 operating systems. We can consolidate servers using virtualisation. Space and power are expensive so we want to cut those costs. Even now when we don’t have money, we can probably justify the cost of a virtualisation project by the reductions in space and power.
- DAS is hot, SAN is not: Yes and No. There’s two ways to look at this. If you know that you have limited and predictable storage requirements then DAS is the way to go. However, if you need large scale or unlimited growth storage then SAN is the only choice. I’ve looked at the cost per GB of DAS VS SAN. DAS for small amounts is cheaper for the purchase but more expensive per GB than SAN as time goes by. DAS also eliminates server flexibility.
- Virtual infrastructure is HOT. Virtual machines are NOT: 100% agreed. I think I read somewhere that Gartner said that the real challenge with virtualisation is the management of it. With virtualisation everything becomes intangible. Virtual machines are files on a disk. Virtual networks are settings in software. This is why we went with Hyper-V. Virtual Machine Manager 2008 gives fantastic centralised management. It ties in with OpsMgr 2007 SP1 to give us top-bottom and cradle-grave management of the entire network, regardless of whether I’m dealing with virtual machines, virtual networks, physical machines, services, applications, hardware, storage or physical networking.
- Dynamic Core Utilisation: The days of one application – one processor are over. Virtualisation has made sure of that. Are you one of those people who has a dedicated anti-virus server, a dedicated WSUS server, etc? You should have a look at virtualisation now. MS is planning some really cool improvements with Core Parking to take this to the next level so that unused loads don’t consume resources. VMware does a nice job with VI3 too.