Customers often approach "the cloud" as if it were some big-box store where IaaS (infrastructure as a service) can be purchased as a commodity. That may be marginally true for simple backup or for dev and test, where the risk of failure is not a big deal. But companies that wish to stand up mission-critical production applications on, say, Amazon Web Services, rapidly discover they need to become experts in the intricacies of the platform.
RightScale, an independent SaaS provider, shoulders much of the burden of building out enterprise-class server deployments for IaaS. It delivers a top layer of management and automation in the form of templates for Amazon, Rackspace, SoftLayer, and other providers, as well as for private cloud infrastructure built on technology from Citrix, Eucalyptus, and OpenStack. Best practices are built right in.
[ In the data center today, the action is in the private cloud. InfoWorld's experts take you through what you need to know to do it right in our "Private Cloud Deep Dive" PDF special report. | Also check out our "Cloud Security Deep Dive," our "Cloud Storage Deep Dive," and our "Cloud Services Deep Dive." ]
InfoWorld executive editor Doug Dineley and I recently spoke with RightScale CEO Michael Crandell, who has a unique perch from which to observe the cloud industry, both on the provider side and the customer side, where he can claim Comcast, PBS, Zynga, and many other big players as part of his customer base. The following is an edited version of that conversation. We were particularly intrigued by Crandell's take on the entry of HP and Microsoft into the IaaS business.
First, we asked him to explain why customers should go to RightScale instead of going directly to an IaaS provider.
Eric Knorr: Say I just want to manage virtual machines on Amazon. What value does RightScale bring to the equation?
Michael Crandell: The first piece is really the automation. There are apps that are amazing in terms of sys admins and servers -- that are able to manage literally thousands of servers through automation, autoscaling, auto-failover, things like that.
RightScale is sort of the big eye in the sky monitoring on these servers. We might detect, for example, that the CPU load had gone up on a bank of app servers. If it went too high for too long across too many machines, let's launch some new app servers. Well, you're now launching new app servers into a particular cluster or deployment that might be running -- if you take Zynga as an example -- Cityville or Farmville.
For the sake of this example, we're just scaling the app servers, not the database layer. So how do the app servers when they launch know how to fold themselves in and configure themselves to be part of the rest of this whole system? That's the dynamic configuration we provide. We use a particular piece of IP that we created for that; we call it a Server Template.
A Server Template in simple terms is still a machine image at the base, reduced down to the plain vanilla OS and a few super common utilities. And then everything that loads to make that server whatever it is in life happens dynamically at launch time.
Doug Dineley: Then the server template is not really a template. It's a set of scripts.
Crandell: It's a set of scripts, the one difference being that you can abstract out variables from the scripts and pass them in at launch time. So in that sense it is a template for, say, an app server -- and you can pass in a launch time, where is the database, what's the host's name, where are the other app servers, where's the load balancer -- and RightScale can dynamically figure that stuff and pass it in. In that way you can make a model of an app server that you can use all over the place.