Although some IT futurists are already talking about the all-cloud enterprise, the industry as a whole is not quite ready for that yet.
At the moment, the cloud is viewed mostly as an add-on to in-house data center infrastructure. If demand is pushing internal resources to the limit, it’s nice to have virtually limitless scale to fall back on. This approach is called “bursting,” according to F5 CTO Karl Triebes. But to get there you need a high degree of integration and compatibility between in-house and cloud systems — in short, you need the ability to “switch on” the cloud at a moment’s notice and instigate a smooth transition of data and applications from one to another.
As you can imagine, this is easier said than done. A key component in all this is a robust wide area network (WAN) capable of handling the high data throughput necessary to provide a seamless experience for users. It would be easy if all it took was a WAN optimization device and a new data management stack, but Triebes says this is not enough. One drawback of standard optimization systems is that they require specific locations for both the central data center and the off-site location. Optimization is a step in the right direction, but it has to be incorporated within a revamped application delivery architecture, one that accommodates not only virtualized applications in the cloud, but all of the networking and routing possibilities between the cloud and data center.
F5, of course, makes its living revamping network architectures to handle the fluid data dynamics of virtualization and the cloud. However, a handful of start-ups is looking to consolidate those architectures down to the appliance level, promising a plug-in cloud model with WAN optimization as a central component. One of the latest is Cirtas, which just came out with the BlueJet cloud storage controller that provides local solid-state and hard disk cache with an integrated optimization system tied to Iron Mountain’s Archive Services or Amazon’s S3 service. In this way, the device provides ready access to critical data and sends compressed, deduped secondary data to the cloud as needed.
Another approach is to place the optimization service on the cloud itself. Aryaka Networks is one of a string of new companies that wants to do away with the on-site optimization appliance in favor of WAN Optimization as a Service (I haven’t seen the WOaaS acronym yet, although WaaS is already taken by Wide Area Application Services). The company claims it can speed up applications by 150 percent and can be called up as needed, unlike traditional systems which hit both capital and operating budgets even under minimal use.
Optimization alone isn’t the answer to smooth-running cloud operations. Cavium Networks is pitching its new OCTEON II CN63XX processor as the key to an intelligent network across the board. Intended for everything from security and network/storage compression applications to WAN optimization and TCP/IP and iSCSI offload environments, the device mixes DPI functions with dual 10 GbE interfaces on a single PCIe 2.0 format. It also supports non-volatile RAM for SSD applications. A smooth flow of data between the data center and the cloud is only as good as the legacy infrastructure’s ability to capture and process it.
This probably won’t be the last we’ll hear of solutions and techniques designed to make the cloud better, faster and more productive. As the most recent of a long line of IT technology advancements, the cloud is only in the initial phase of development.
At the moment, getting onto the cloud is relatively easy. Making it sing will take a little more work.