Blade Watch

In the IT world you’re typically measured by three main things:

1. Server provisioning – how long does it take to install a server for the end user?
2. Outage percentages – how many days downtime have we had?
3. Service Level Agreement benchmarks – how many calls were answered in the 15 minute response time?

Why the server provisioning, because it can actually become quite an emotional exercise. Let’s take an example, if I today want a server and I work for a large company, I’ll typically fill in the paperwork, get it signed off and then the following chain of events happens:

• IT purchase a server
• The server arrives to the loading bay a few days (even weeks later)
• The data center team allocate space
• The hardware team rack the server
• The network team then allocate a network port and ip address
• The patching team provide a network cable
• The server team (windows/unix) install the operating system

At this point I have a working server from an infrastructure standpoint, it’s got windows, I can ping it, but I can’t log on to it.

• The server is prepared for me – application code loaded, access and shares created
• Monitoring is enabled

The lead time differs from business to business, industry to industry, the longer the ‘delay’ in delivering my server the higher the cost, perceived or real. The main challenges in server deployments are caused by a lack of planning, human error, or often a change in requirements.

• Lack of planning – but you said the application worked on Windows 2003 – “no it’s verified on Windows 2000 SP4 – 2003 isn’t supported” – re-install Windows please
• Human error – that the business team asked for the development network when they meant staging, means new ip address, new network port and network cable
• Change in requirements – that we actually need the servers to be clustered – great, but that means we need SAN storage allocated, and then the Microsoft clustering software installed

With this in mind, some of the banks are switching to a pre-provisioning model to improve their ability to have an on-demand infrastructure, one in which the IT can provision a server in minutes not weeks. IT will buy for example 500 blades, and sell ‘the service’ or ‘the asset’ back to the business on-demand.

This though is where virtualization comes into play, both in terms of the server as well as the storage and the network. The ability to create an instance, allocate it an image and have it working in minutes is where you want to be. Where I log into the site, select:

Windows 2003 with IIS, 1024MB RAM, 2 virtual processors, 20GB for C, 20GB for D, click the submit box, my boss approves the server, the cost, and the build process kicks of the imaging and configuration process, then emails me when complete – on demand in it’s true sense.

Related posts:

1. Interviewing a server build engineer For the second time, I had a conversation with a...
2. 3PAR enhances their platform to streamline provisioning CNN Money 3PAR® (NYSE: PAR), the leading global provider of...
3. Interviewing a server guy (part 2 of phased approach) So we left Mike wanting to upgrade all the servers...

Related posts brought to you by Yet Another Related Posts Plugin.