Since our previous Buyer's Guide of nearly two years ago that reviewed Stratus, the company has been busy. In 2009, it moved to Nehalem quad cores across its ftServer platform (ft = Fault Tolerant) with the ftServer 2600, 4500 and 6300. In September 2010, it launched the ftServer 6310 hex-core (six cores) Nehalem, with increased amounts of memory to suit VMware Tier 1 application support in the data center.
"Hardware fault tolerance for running VMware (and Hyper V) allows for symmetric multi-processing (SMP), eliminates the need to cluster for availability, reduces software licensing, and prevents downtime and VM restarts caused by platform failure," said Denny Lane, director, product management & marketing, Stratus Technologies.
On the low-end ftServer 2600, the company added support for Windows Server 2008 Enterprise Edition in addition to Standard Edition to establish it as a possible home for SMBs adopting virtualization and for remote office/remote branch applications.
The biggest change in the past couple of years, though, is the addition of Status Avance high availability software. This is the company's first software-only product and the first one to run on generic x86 servers. Avance can create a high-availability virtualization platform from two standard servers. Much like the ftServer system software, it monitors, diagnoses and prevents failure before it causes system downtime. Both products "call home" to Stratus customer service to report issues that the system itself cannot auto-close.
|Stratus' ftServer System|
"We have come to realize the value that customers attach to this automated uptime middleware and proactive availability monitoring and management services because now they don't have to worry about Stratus platforms," said Lane. "Uptime assurance is what customers get from us."
Meet the Stratus ftServer Family
The Stratus ftServer family is the mainstream product set. There are currently three ftServer models, all of which are configurable Intel Xeon-based servers that support Windows Server and Red Hat Enterprise Linux operating systems, as well as Hyper V and VMware vSphere. With the stress on uptime, it's no surprise that most customers run mission-critical applications. According to Lane, this line operates at 99.99976 percent uptime. He added that the installed base of ftServers had an average of 62 seconds of downtime in 2010.
The ftServer 2600 is an entry-level quad-core single socket server primarily for department, SMB and remote office/branch offices (ROBOs). An ftServer 2600 ROBO package consists of: one ftServer 2600 system with 16 GB memory; 12 146GB 15K rpm disks; and one Microsoft 2008 Windows Enterprise Edition OS. Lane said a comparable HP system would be priced 27 percent higher.
At the middle of the range is the most popular Stratus model, the ftServer 4500. It can be either a one- or two-socket quad-core machine for midsized companies or departments running core data center applications or databases, a fault-tolerant computing resource in a virtualization server pool, a cluster replacement or used for server consolidation. The ftServer 4500 includes 2GHz E5504 Intel quad-core Xeon processors, and the Intel QuickPath Architecture for high-speed connections between microprocessors and external memory as well as between microprocessors and the I/O hub. A one-socket version has 4GB to 48GB of memory, and a 2-socket 4500 has double that amount. Up to 16 SAS 2.5" drives of 15K (73 and 146 GB) and 7.2K (500GB) are available. The Stratus ftServer 4500 starting price is approximately $24,000.
According to Lane, the ftServer 4500 typically competes against x86 server clusters from HP, Dell and IBM. "Increasingly VMware virtualization software is assumed to provide adequate uptime; what is not factored in is the need to over-configure, the added configuration and management complexity, and failover," said Lane. "VMware FT is not even close to hardware-based fault tolerance by any metric."
At the top of the line is the ftServer 6310. This two-socket hex-core model is aimed at major enterprises and data centers for use with mission-critical applications and high-volume transaction processing.
"Running Oracle on an x86 platform in a VMware infrastructure can save hundreds of thousands of dollars," said Lane. Note that an ftServer 6300 is still available with a quad-core processor with a higher clock speed than the 2600 or 4500.
Additionally, the V Series version of the ftServer line is sold to customers that want to migrate their Stratus VOS-based applications to and from Stratus' legacy Continuum servers.
How does Stratus view the server marketplace? As the number of virtual servers has surpassed physical servers for new applications, IT management needs higher-end, more powerful servers to support increased application loads on fewer servers, said Lane. "Impact from a platform outage increases the need for improved uptime assurance, such as Stratus systems provide," he said. "The same holds true as companies look to build out their private and hybrid clouds."
He laid out some math. Instead of comparing the raw hardware prices, companies should be factoring in the costs of downtime, which will be more in some organizations than others. Forrester Research pegs the average cost of one hour of downtime at $150,000 for the typical enterprise. That, of course, applies to large and IT-centric organizations. The closer the cost of downtime comes to that figure, the more appeal for a platform such as the ftServer. "Understand your cost of, and tolerance for, downtime," said Lane. "Most IT shops have no clue what the impact of downtime is on their organizations. At a minimum, an hour of downtime will cost tens of thousands of dollars at any moderately sized business." To put its money where its mouth is, Stratus is forming partnerships to link its fault-tolerant servers with critical applications, such as large-scale Oracle and Microsoft SQL Server, Exchange, SharePoint and Hyper-V implementations. The company will be guaranteeing 100 percent uptime for an ftServer running Windows or VMware, with a $50,000 payout for any downtime caused by its products or the OSs.
Drew Robb is a freelance writer specializing in technology and engineering. Currently living in California, he is originally from Scotland, where he received a degree in geology and geography from the University of Strathclyde. He is the author of Server Disk Management in a Windows Environment (CRC Press).