I've been following the development of the Inter-Cloud concept with interest over the past many months. Popularized by Cisco (the network is the cloud), Inter-Cloud is described variously as a “federation of clouds based on open standards,” an “elastic mesh of on demand processing power deployed across multiple data centers” (in this post by Greg Ness) or simply a “cloud of clouds.”
The posts I have read on the topic deal with issues such as standards for network protocols, virtual machine representation, and identity/security. The vision, once these details have been worked out, is a completely ubiquitous computing environment that seamlessly blends all forms of clouds and their physical instantiations into what seems to me to be a single, massively parallel cloud computing environment.
But as awesome as that vision is, it won’t solve the pesky interoperability challenge of the business applications that run on these linked clouds. And assuming we want to do something more interesting than host websites, email and excess storage, existing and new apps need to migrate to the cloud. In the virtual layer cake in the sky, we talk about Infrastructure as a Service, Platform as a Service and Software as a Service. But too many cloud "taxonomy" and “ecosystem” charts leave out the critical element of application integration. It’s time for integration to take its rightful place on the layer cake!
What Inter-Cloud does for cloud computing (compute as a utility), integration does for SaaS (applications as a utility). Inter-Cloud may make cloud computing environments interoperable but it is SaaS integration platforms (think "cloud middleware") that make SaaS (and on prem) applications interoperable.
I blog in another post about "the case for cloud middleware." We have the opportunity with the SaaS industry to not repeat the mistakes we made in the enterprise era with hardcoded, brittle, essentially dead integration. We have the opportunity with cloud middleware to implement living, breathing, scalable interfaces that will enable and drive the growth of SaaS and cloud computing. After all, if business apps don’t make the transition to the cloud, the future of SaaS will certainly be in peril and the full potential of cloud computing will not be realized.
The industry does not need a complete spec for portability of VM instances, but rather a secure transaction broker spec for simple failover.
The insurance industry is all over the creation of a "blind trusted pool" of API brokers between clouds that would allow erastwhile competitors to backstop each other's failures and outages without compromising user security. They are concentrating on transactions, storage proxies, and session preservation. In other words, short, long, and very long object lifetimes that can be shipped out to an unknown (yet trusted) third party in case of an unreachable host.
Posted by: abm | July 21, 2009 at 12:27 PM