You could do the same thing with a piece of blog software, say, WordPress, and abstract out the name of the blog; the graphic that it loads with; the admin name, password, and email; stuff like that. Abstract that out and you get a reusable template, if you will, of a blog server that you can run. And it's a very generalizable concept. I've given two examples, but you can imagine.
On top of that construct we decided to build a marketplace. Server templates are cool; they're reusable. We open this up in two ways: One is for partners to publish and the other is for companies to share -- within the company or outside the company. So we actually have a community at RightScale. We have about 55,000 people who have signed up.
To give some examples of ISV partners, we have Riverbed with their for-pay load balancer and Zend, the PHP accelerator company has built a whole scalable cluster.
Knorr: Sounds like a very open platform. How do you control who does what?
An important part of RightScale is governance and access control over what users are able to do. Imagine I worked for a company and on my smartphone I could literally bring up a dozen different accounts with many cloud credentials and cloud resources and probably access to thousands of servers -- and I could show it to you and press the wrong button. So we provide access control over what users are able to do according to what roles they play and what permissions they have.
Then there's auditing and logging. As you know, Amazon is a world where if you start your log files on the local disk and if you're not using EBS and you terminate the server -- it's gone. So you ask: What happened on that 137th server we launched last Monday for that batch process? It's gone. So we persist logs, we persist audit entries -- which by our definition are like stages of important events -- lifecycle stuff. And then cost accounting. We have a lot of visibility into what people are running.
We can report back to them that this deployment, which runs, say, your business intelligence app and cost X amount in terms of compute, storage, and networking last month. We do some predictions on that with our dashboard, too. Then we deliver that as a CSV file that you can take and pump into your accounting system and get some granularity around what your spend is.
Dineley: The business side must like that since Amazon pricing can be a very black box.
Crandell: Also, there's a piece of this that we call the IT vending machine. It's a cute name for self-service IT. The notion is that your IT department might get into the heart of this, build some workloads that they might want to provide to business users, then wrap a simple interface that says: Push this button, pick the cloud you want, enter your name and whatever variables. Then they know what workload is going to run there. What's exciting to us is that they deliver the agility of the cloud with control and governance.
Knorr: But as you say, IT still has to set up the vending machine. To what degree can you insulate IT from having to know the details of implementing infrastructure on Amazon? From what we hear, Amazon customers are often surprised at the level of expertise they need to acquire to make things work.
Crandell: We do a fair amount of insulation. Look, any of us could download MySQL, stick it on a server on Amazon, and run it. But that's not what people want. They want a resilient, high-performance, backed-up, failover-ready installation of MySQL because it's their data.
If you're going to build that, that's a whole different potato. You have to set up replication, snapshots, you have to have a runbook for what happens when something goes wrong. All of that has been baked into a MySQL set of templates that RighScale happens to provide to paying customers, and it has all of that operational goodness in there -- which is frankly becoming more and more of the equation.