New Tools in the War Against Latency

Arthur Cole

For much of its history, IT has been tasked primarily with maintaining sufficient resources to meet data application and data needs.

But now that virtualization and the cloud have taken firm hold of the infrastructure and resources are available at the click of a mouse, IT's focus has shifted toward the reduction of latency. This is a nod to the fact that even though server, storage and other resources are accessible, they are not necessarily close. Data will likely traverse a patchwork of internal and external network infrastructure as it moves from source to destination, with varying amounts of available bandwidth between each hop.

This problem has become particularly acute for banking and other financial applications, as even a few microseconds between request and fulfillment can cost millions. AdvancedIO Systems is targeting its new V5031 10 GbE card at the financial sector, boosting performance to real-time levels with the new 28 nm Stratix V FPGA from Altera. The card features four optical 10 GbE ports, plus a range of advanced memory configurations that put trading systems, risk controls and other functions next to the network where they can be implemented more quickly. It also features the expressXG development framework for fast deployment of network, memory and host interfaces.

Storage is also a prime target when it comes to cutting latency, which is why interest in Flash technology has grown in the past few years. Texas Memory Systems recently unveiled a new version of its RamSan server cache system that features a 450 GB eMLC PCIe board tied to the CacheWorks management stack from NEVEX Virtual Technologies. The combo provides direct storage connection to memory resources via the PCIe interface, coupled with application performance and a Windows memory cache, to boost application I/O five-fold.

Still, even the lowest latency infrastructure is of limited benefit once you start to push applications onto the cloud. To keep the data flowing, you'll need one of the new generation of cloud-optimized acceleration platforms, such as the Terra Alta system from Akamai Technologies. The system provides a number of tools, such as edge management and load balancing, DNS mapping and dynamic page caching, to maintain application performance as resources fan out across internal, external and even mobile environments.

In certain applications, however, latency isn't just a nuisance but a downright killer. VDI, for instance, has fought a never-ending battle against latency as it strives to match the performance of traditional desktop infrastructure. PC-over-IP developer Teradici is hoping to close that gap with a new server offload card optimized for VMware ESXi 5.0 environments. The APEX 2800 is designed to take the heat off of CPUs caused by the deployment of virtual GPUs under ESXi 5.0. The device also monitors all VMs on its host server and utilizes VMware DRS and vMotion to consolidate users onto minimal hardware footprints without having to rearchitect existing VDI environments.

Latency, then, is not necessarily about speed, but about efficiency. The faster data and applications can traverse IT infrastructure, the more resources become available to handle additional loads. Users will appreciate the faster response while the front office sees an improved bottom line. Everybody wins.

Add Comment      Leave a comment on this blog post

Post a comment





(Maximum characters: 1200). You have 1200 characters left.




Subscribe to our Newsletters

Sign up now and get the best business technology insights direct to your inbox.