Super Reliable Systems At The Ready

Just don't call it clustering.

"We are the alternative to clustering," says John Leonard, worldwide public safety manager at Stratus Technologies. "We never use the word 'cluster.'"

The Maynard, Mass.-based company works with a number of state and federal agencies in implementing a fault-tolerant system that is packaged in one server--one that reportedly averages fewer than four minutes of downtime per year. That's better than 99.993 percent availability.

According to Leonard, Stratus developed a system that duplicates each component in the device. That means two motherboards, two memory boards, two disk drives, two I/O slots and so on. "Everything is doubled up except for the clock set, so there is one clock that runs the entire system," he says.

But the unique part is it only needs one operating system, which is Microsoft Advanced Server 2000. That's what VARs say is so appealing about this device. Needless to say, the prevalence of the Microsoft OS, combined with Stratus' fault-tolerant design, help make the system easier to administer. That's an attractive pitch to government organizations, which do not have the dollars to hire top IT expertise.

"Customers get high availability at an easier administration level. That is pretty critical when you are answering 911 calls," says Cy Cole, vice president of sales and marketing at Open Software Solutions. "We have had fault-tolerant servers before, but with proprietary operating systems. As far as I am aware, Stratus is the only company that offers fault-tolerant server systems that use Microsoft's operating system."

The system has no failover capability, in which one server is active and another is passive until a glitch brings the active system down and the second one kicks in. The redundant parts in Stratus' ftServer are running continuously. The system has diagnostics and utilities that flash green, yellow and red signals to highlight different levels of failure. If both sides are green, that means both component entities are working. If it flashes yellow on one side and red on the other, that is a warning one side is failing.

At that point, the IT manager can pull out the 1.75-inch unit, while the call-home feature has notified a Stratus worldwide customer assistant center. "The next day, FedEx will deliver a replacement unit. It literally slides in and slides out," Leonard says.

Through all that, the application continues to run. The solution is popular among banking, enterprise-telephony and manufacturing sectors, as well as fire departments and EMS services, according to Leonard.

Stratus executives say that research firm IDC identifies several options when it comes to setting up server environments for data availability: a standalone server setup that either uses disks for mirroring or RAID; a type of backup configuration where data is copied from one server to another; a cluster configuration for automatic failover; and a single box with built-in redundancy for high availability. Stratus' product falls into the last category.

Last quarter, the company closed roughly 48 deals in the public-safety sector. Police officials, in particular, would take a shine to this kind of product, Leonard notes. They have applications that demand unencumbered uptime, but that do not demand a lot of technical expertise.

"No cop is interested in being a computer guy," he says.