The up.time IT Systems Management Blog

As most of you are probably aware, the ‘Cloud‘ is garnering a lot of press coverage; and is already being dismissed for its vagueness and marketing-speak.  The cloud is currently being pitched as a panacea to all our enterprise computing problems — high availability, unlimited capacity, instant provisioning, better end user experience, and seamless integration with existing enterprise infrastructure.  Which given time, will probably all come true, just like autonomic computing… (remember that?)  More on the ‘cloud’ a little later.

As a vendor in the system’s management space, we continually have to fight the urge to leap ahead into the ethereal world of marketers or the analyst cycle of “peak of inflated expectations“.  There is so much cool stuff out there, it’s difficult to not get caught up in the hype.  What we have to remember is that the press and analysts (and I am somebody that does get value from analysts) live in a Fortune 500 world, where the biggest and most complex environments exist.  In these companies, there are lots of top-notch IT people with many resources for researching and testing new technology and they can experiment with niche vendor solutions.

Our challenge is to take a step back and identify what problems is the rest of the market experiencing, or, can we sell a lot more of something less complicated.  I equate this to our exposure in the virtualization space.  VMware has been out for many years now and the technology is mature.  Over the next 18 months some really interesting software/hardware will be emerging, however, the majority of the market just isn’t going to be adopting the new bells and whistles, there’s going to be lag-time.  It’s as if people are being marketed a Ferrari when they’ve only just come out of the cave and discovered fire.

In a recent Information Week (PDF) survey on virtualization, there’s some great discussion on VM sprawl, VMotion Sickness, and expenditure on management tooling.  One interesting point, however, is that the adoption of management tooling is much slower than expected; lots of people are still doing things manually, or with minimal management automation.  

Which brings us back to the ‘cloud.’  The concept is just too ethereal for the IT staff that need to keep systems running.  The management tooling just does not exist for cloud management, and who in their right mind is going to do a production deployment without active monitoring.  Look at the largest cloud vendors (Google Apps/Gmail/Amazon S3, EC2), they have had considerable outages and performance pain in the recent past; and these are vendors with incredible technical expertise.  Could you imagine have dynamic compute allocation and then getting a bill for 10,000 hours of CPU time… and not even know about it in advance!

To me (and slap me for my contrived short-sightedness), the current IT world has a lot of other problems to solve before leaping into the ether. 

So what have our users told us they’ll be working on next year (vis a vis virtualization)?

• There’s a lot more widespread of adoption of Xen and Solaris zones/containers.  Xen is increasing because it’s part of the standard install options in most recent Linux distributions.
• VDI is an active project in a few of our largest customers, VD-what? in all others.
• DR using VMware is an active project in most customers that have had VMware for at least 18 months.
• Virtualized storage is starting to become an issue for larger SAN customers (NetApp virtual heads).  I’m excited by this problem because it’s pretty much going to be difficult to map virtual instances across virtual networks onto virtual storage pools (does that mean you have virtual performance problems?)
• Storage management within a virtual environment is a large problem (e.g. why did all that space just disappear?)