
At VMware 2011, vendors and customers alike are recognizing the critical need for VMware acceleration to address the virtual crawl/stall phenomenon. Industry surveys show that on average, most VMware virtualization projects have only virtualized 30% of data center servers (file, print, web server apps). Performance issues have caused most companies to “stall” the virtualization of the core business critical applications that make up the other 70% (OLTP online, transaction processing systems with databases, emails, collaborative computing apps).
While headlines like EMC Achieves 1,000,000 IO/s and 10 GB/s In VMware vSphere 5 Environment make great sound bites, performance alone is not an effective metric. Instead, customers need to ask: performance at what cost? Like most issues, given an unlimited budget and IT resources, companies could easily add performance to their VMware environment. But if adding a million IOPS costs a million dollars, it is not a feasible or practical solution. Dollars per IOP is a better metric.
To overcome the virtualization crawl/sprawl issue, companies need an acceleration solution that is cost-effective, flexible, and scalable. A solution that delivers 80,000 sustained random IOPS for under $30K (like ViSX G3) is affordable to any customer budget, (about the cost of a fully configured server). Being able to add performance to the VMs that need it (made possible with Networked Performance Flash™ ) without limiting performance to a single host machine—adds a critical level of flexibility. Plus with vSphere 5, multiple ViSX G3s can be clustered together for a scalable solution.
More and more vendors and customers are recognizing that VMware performance acceleration is a critical need in today’s data center. However, companies need to choose a performance solution that is cost effective and practical for wide adoption.
— Posted on September 13, 2011
I am pleased to announce the launch of the Astute Networks ViSX G3 solutions for VMware virtualized environments. This product will fill a critical need in the market for a simple, cost-effective way to boost the performance of virtual machines (VMs) and virtualized desktop infrastructures (VDI). Analysts and customers alike are singing its praises. It is the first product on the market to solve the “virtual stall” problem caused by performance bottlenecks. I predict that IT managers will recognize its value immediately and make it a standard part of every virtualization environment.
In many companies, virtualization implementations got off to a roaring start—dozens of file/print, web, and other low-IO servers were consolidated onto a few host machines. You not only had fewer physical servers to manage, you could move operations, manage multiple operating systems, and allocate resources quickly and easily. And then virtualization stall set in …
Virtualization of the data center came to a screeching halt as soon as you started to virtualize high IO applications and databases. You soon found out that Exchange, SharePoint, Oracle, and SAP were too slow in virtual environments. Ironically, these are the very servers that need the consolidation, flexibility and availability of virtualization most. Unfortunately, they also create an enormous need for random IO that common virtualized storage hardware just can’t deliver.
Desktop virtualization efforts suffered a similar fate. The early morning boot storm caused by employees all booting up at the same time typically brought performance to its knees. And caused an early end to VDI projects. The reality soon became clear—VM performance is a critical requirement for successful virtualization.
You may have tried adding spinning disks. Physical disk storage is optimized for capacity, but cannot handle the random IO requirements of a virtual environment with any efficiency. Adding conventional disks to get performance is like putting out a large fire with a bucket brigade. It may improve the situation, but it’s not efficient or cost-effective enough to be a practical solution.
Adding Storage Area Network (SAN) or direct attached storage (DAS) flash aren’t practical either. SAN flash is tied to legacy architectures designed for rotating media and limits flash performance. Both DAS and internal PCI storage flash lock performance to a specific physical machine - a significant limitation in virtual environments where you need the flexibility to allocate performance as it’s needed by a given VM. Other flash solutions don’t support VMware features completely either.
In contrast, ViSX G3 is a simple and cost-effective VMware Ready certified solution that boosts performance of VMs in your environment. Its innovative Networked Performance Flash&trade architecture lets you simply plug it into your existing infrastructure via a standard Ethernet port and lets you add performance to a virtual environment where you need it. It uses the patented and unique DataPump Engine™ processor to accelerate ViSX G3 solution TCP/IP network and iSCSI storage traffic and deliver the equivalent of hundreds of high performance enterprise hard disks in just three units of rack space. For a cost of a few hundred dollars per VM, it lets you provision performance with the same flexibility and control as you provision capacity in your virtual environment.
With a simple, cost effective way to boost the performance of virtual servers and desktops, ViSX G3 solutions let you realize all of the benefits promised by virtualization:
We are excited about the possibilities this powerful solution opens for our customers. I look forward to sharing my insights on the latest news and trends in virtualization with you through this blog.
— Posted on August 15, 2011