Astute Networks Fixes Virtualisation ‘Crawl And Stall’

Astute’s ViSX G3 uses network performance acceleration to clear congestion in VMware environments

Astute Networks, which specialises in NAND flash-based appliances for virtualised infrastructures, on 15 August introduced a network performance accelerator specifically for VMware environments designed to clear out workloads faster via more efficient use of resources.

Astute ViSX G3’s first job is to enable efficient adoption of server and desktop virtualisation within a network. But it also can benefit private cloud systems because it features sustained random I/O performance – using optimised flash memory – to virtual machines over standard local area networks.

Non-essential applications

“A lot of network administrators are quite familiar with the catchphrase ‘crawl and stall’, which happens when virtualised networks get overloaded with server storage,” Astute Senior Vice President of Marketing Jay Kramer told eWEEK. “This is a purpose-built hardware and software package that alleviates all of this.”

Industry analysts, such as Yankee Group and Forrester, have reported that less than 15 percent of data centres have virtualised their main business applications, Kramer said. Most enterprises have virtualised only non-essential applications primarily because of the virtual stall problem and performance concerns, he said.

In many legacy systems, databases, virtual desktops, and cloud computing environments generate more network and storage I/O than virtual environments and storage can handle, which is the main cause of “crawl and stall”. Normally, those frontline applications would be prime candidates for being virtualised, but many CTOs and data centre managers are hesistant to make the move if the older system they currently use is working well enough.

Virtualised storage has inherent bottlenecks that must be solved to avoid “crawl and stall”. The basic problem is that data gets scrambled as though in a blender when it travels from servers through pipelines to a hypervisor and then into storage containers. Reassembling increasing amounts of data wears heavily on a conventional system because it takes extra time – at ultimately higher cost – for unoptimised systems to straighten all the bits out and get them put back together so they can be used.

Add in all the virtual machines being created, and this increases the density of the VM farm, presenting even more control issues. As a result, networks are often too slow to support SAP, Oracle, and SQL databases; Microsoft Exchange and SharePoint service levels and user mailbox loads; and backing up or recovering virtualised data stores.

Acceleration

The ViSX G3 box, which will be demonstrated at VMworld in Las Vegas from 29 August to 1 September, consists of of enterprise flash memory, the company’s own DataPump Engine processor, and some other custom hardware and software. Each ViSX G3 supports 64 virtualised host servers and their virtual machines with dedicated I/O.

Kramer said the ViSX G3 also delivers fully offloaded and accelerated network traffic (TCP/IP), virtualised data store traffic (iSCSI), and multi-level RAID protection.

The ViSX G3 is available in three models: ViSX G3 1200 is priced at $29,000 (£18,000), the ViSX G3 2400 is $59,000 and the ViSX G3 4800 is $94,000.