Undifferentiated Heavy Lifting

April 21st, 2008 javier

This phrase was used close to a dozen times by Werner Vogels, CTO of Amazon.com at his recent keynote at the MySQL conference. Werner used it to describe the day to day tasks of most web operations teams… tasks like racking boxes, configuring routers, and installing software. He mentioned ops teams at Amazon got to spending 70% of their time in this mode, and this was one of the main catalysts for developing infrastructure that brings us S3, SQS, EC2, and more.

Removing this “undifferentiated heavy lifting” (hereto referred as UHL) from your cycles is supposed to free operations to spend more time actually operating. As I was listening to the talk, I wondered how many folks believe there is some competitive advantage in UHL? Certainly choice of hardware, network architecture, data center setup all fall under UHL and can mean the difference between success at scale versus utter failure. Nonetheless, the argument is that you shouldn’t reinvent the wheel and should take advantage from those providers (Amazon in this case) who do the UHL for you, right? Well, a lot of people certainly think so based on how much press and use is being directed towards the cloud.

This begs raises the question: “If racking boxes, configuring OS’s and so forth is UHL today, what will be UHL tomorrow?” That question is material to companies in the systems management market because getting caught on the wrong side of UHL means your future is (as our favorite magic 8 ball would say) “Outlook Not So Good”.

Luckily, UHL today is mostly about the pain associated with hardware and network provisioning and configuration. These problems are bounded just enough to make it feasible for someone to simply delegate them to a cloud provider (as many already have). Of course, the implications to management vendors focused on managing UHL tasks are not pleasant if you buy the idea that most applications will move into this sort of an environment. If you buy this vision, then conceivably the future will require one GIANT Tivoli license for the One-Cloud-Provider-To-Rule-Them-All, right? Heh, not quite, but it’s fun to think about it that way!

The reality is that this trend is forcing both service providers as well as application developers to rethink their operation strategy. Providers want to be more like clouds, developers want to run inside them.

Why rethink their ops strategy? Because the stuff that sits above the UHL layer… the middleware, the databases, and most importantly, the code that makes up a given application present the most daunting challenges due to their complexity. This leaves a meaty management problem yet to be solved. One that has more to do with managing complex software stacks with components which might reside inside or outside of the cloud and with an ‘elasticity’ (to borrow another one of Werner’s terms) which demands equal agility in the management layer than what is found in the software stack.

This is where the next big wave of innovation will happen in the management space, and we’re excited to be a part of it. Come visit our booth at the Web 2.0 Expo this week, and you’ll see what I’m talking about.

Entry Filed under: Events, IT Industry, Javiers Blog