Amazon’s datacenter in Tokyo may have been one of the worst-kept secrets in cloud computing and finally it’s public! We’ve been operating there to support customers for a while and can now turn on our dashboard support for the EC2 AP-Tokyo region for everyone. We already have a number of RightImages installed and we’ll be publishing a full complement of RightImages and ServerTemplates with our next release in a couple of weeks. We also already have employees in Japan and RightScale K.K is being formed, so please contact us with any questions. All this is great news for us and our Japanese customers, many of whom have been operating out of US regions thus far.
When we announce full support for a new cloud this implies a significant commitment from our end (each EC2 region is its own independent cloud and the only thing that is shared across all EC2 regions are the account credentials). While RightScale started as a single small instance back in 2006 it is now a global distributed system. The best way to think about it is that it provides all the life support that servers need to function and be managed efficiently in a growing set of clouds around the globe. We are in the midst of a year-long engineering effort that allows us to distribute more and more of the components that make up the RightScale platform. This is important because we can move configuration data and communication infrastructure into each cloud to improve performance as well as availability — which ultimately includes multi-cloud business continuity.
I must say that the number of life-support systems we have had to build has come as quite a surprise to me and I believe that this is one of the distinguishing features of the RightScale platform. It turns out that provisioning, configuring, managing, monitoring, and automating servers requires a lot of moving parts. On top of that, our customers expect to see their servers running in different clouds all on one web page. The RightScale life-support system for managing servers consists of many parts:
- when they start to boot, servers need to connect back to the RightScale core to get their configuration information
- during the boot process system packages need to be downloaded for installation or update, this requires redundant in-cloud OS distribution mirrors which we provide; in addition, we provide daily mirror snapshots so it is possible to boot a server with a fixed known mirror content
- during the boot process the ServerTemplate’s content needs to be downloaded, whether RightScripts, RightScript attachments, or configuration data; these also need to be cached redundantly in-cloud, while updates to the origin (svn or git repos in some cases) need to be propagated promptly
- as servers come up they often need information about other servers in the same deployment, for example to locate load balancers, app servers, and/or databases; we support this using tag lookups and server-to-server messaging through RightLink
- once servers are operational they produce monitoring data which needs to be collected, stored, and graphed; similarly they produce syslog logging data which must be collected
Most of these support systems are now located in all the EC2 regions and we are continuing to distribute more of the platform. In time this will enable us to place support systems into larger private clouds as well. But I’ll keep that for a future blog post since right now we’re busy rounding out support in EC2 Japan as well as localizing our dashboard.