One of the discussions that came up again and again was around when, how and why Enterprise business was adopting Cloud Computing. This is always an interesting discussion at Cloud Connect because in the past the thinking went something like this:
Enterprise = Private Cloud = False Cloud = Old Stupid Apps = Stupid People = Waste of Time
A next-level down the thinking was that Enterprise ways of building clouds were too expensive, not designed for elasticity and would be ultimately ineffective for web-scale applications. Hence, building or operating anything called a "Private Cloud" yourself was the quickest route to the IT unemployment line and guaranteed to bring shame and ridicule upon your family.
Let's now fast-forward to 2012....
My podcast co-host Aaron Delp wrote this week about some of the discussions that happened on the InterTubes to in parallel to Cloud Connect:
The highlight of the last week was the "hoopla" to come out of Cloudscaling over their Open Cloud Infrastructure announcement and Chris Hoff's response and then Randy Bias' counter-response. Great reading and Randy provides some good background in the last article!
As this blog title states, I believe Randy Bias is (generally) right. But not for all the reasons he lays out in his announcement and counter-response.
- Randy was right many years ago to state that web-scale apps are different than "traditional" (read: legacy Enterprise) apps, at both the infrastructure and software layers. He wasn't the first to say it, but he said it very loudly because it's what he believed and because it supported his (eventual) business model.
- Randy was right to compare AWS to VCE Vblock as the good vs. bad examples of Cloud Computing. Not because he was accurate about market forecasts or failures (both are running at about $1B annualized run-rates), but because it's hard to drive a religious movement unless you can clearly identify the "bad thing that people should move away from". Some would say the good news is that it's becoming less about good vs. bad and more about an expanding pie.
- Randy was right to (eventually) realize that Enterprises hold most of the buying power in IT and generally don't feel comfortable with running workloads in "shared Cloud/Service Provider" models, so it's important to offer the value of your company to those buyers. They'd also like the flexibility to leverage Cloud/Service Providers in the future.
- Randy was right to claim that trying to merge "legacy IT" (he calls it Virtualization 2.0) with an development/operational environment for web-scale apps (mobile, big data, social media, etc.) is going to be very difficult. It would involve combining architectures that weren't designed to be combined, and combining operational models that are like oil and water.
Of all of those things, I think he'll probably be most right on #4. Change is hard, especially for people. Having a blueprint for doing something different is a good first step to implementing something new/different. But it's already a crowded space with several other companies attempting similar models
Whether or not CloudScaling will be successful in winning customers or if customers will care about the "Open" vs. "Non-Open" debate isn't as important to me. Those issues require the mixing of business execution with religion. I like to see those kept separated, but call me old-fashion.
My kudos to Randy Bias for continuing to push many things in Cloud Computing. I don't always agree with his tactics, but I do think he got several things right at Cloud Connect in 2012. It'll be interesting to see where the industry is a year from now...