Application Acceleration in the Clouds
Acceleration technologies can dramatically improve the performance of emerging cloud, SaaS and other Internet-based content and applications.
Distributed data environments have already proven their worth as enterprises flock to remote storage and other solutions on the cloud. The next step is to refine the distributed application layer to allow employees the flexibility to thrive in increasingly diverse and dynamic work environments.
The holy grail, of course, is to create an experience akin to local infrastructure. Most users will tell you that responses are never as fast once you hit the network, but that may be changing for two reasons: One, application acceleration platforms are become more powerful, and two, user expectations are diminishing as the PC gives way to the mobile device as the client of choice.
For the data center manager, then, the question is not so much whether to deploy an acceleration platform, but where and what kind. The answers, however, will depend very much on your legacy infrastructure.
High-speed environments, particularly those populated with SSDs, will most likely benefit from systems that deliver intelligent caching as close to the application processing infrastructure as possible. LSI's Nytro portfolio, for example, utilizes a Flash- and PCIe-based cache architecture plus management software that tailors itself to various types of applications to boost performance in high-end environments. The line includes the WarpDrive acceleration card that scales from 200 GB to more than 3 TB of MLC or SLC cache, as well as the MegaRAID card for use with SAS-connected DAS solutions.
Still, it's getting very difficult to talk about application acceleration without bringing up the cloud, as most enterprises are looking at all the vast resources out there as more than just a data repository. And in a reversal of sorts, it seems that many cloud-optimized acceleration platforms are starting to target the data center rather than the other way around. Akamai Technologies, for example, has tied its Terra Alta platform to specialized virtual machines that can be deployed on the enterprise edge. This allows the system to extend mapping and other functions into the data center so as to provide a unified acceleration platform across internal and external architectures.
But even physical-layer solutions are starting to show increased scalability to better accommodate cloud infrastructure. For instance, the newest additions to F5's Viprion Application Delivery Networking (ADN) platform, the 4300 blade and 4480 chassis, offer support for 40 GBE networks and dramatically boost both Layer 4 and Layer 7 throughput to enhance application and service performance levels. It also kicks hardware compression to 80 Gbps to more than quadruple acceleration performance in many environments.
Speed isn't everything when it comes to boosting application performance, however. The most effective platforms have the ability to drill into the application itself to determine the appropriate levels of service or whether acceleration is warranted at all. That will be increasingly difficult as cloud architectures gain a greater share of enterprise data environments and users start demanding increased collaboration capabilities and the ability to mix and match not just applications, but their components as well.