HP Taps Marathon Technologies for Availability Offering
Behind the scenes, Marathon Technologies Corp. has quietly sold 500 Windows NT-based systems that the company claims provides 99.999 percent availability.
Several of the largest hardware manufacturers made news during the first half of this year by announcing 99.9 percent uptime agreements for Windows NT. But behind the scenes, Marathon Technologies Corp. (www.marathontechnologies.com) has quietly sold 500 Windows NT-based systems that the company claims provides 99.999 percent availability.
Marathon recently garnered more attention by signing an OEM agreement with Hewlett-Packard Co. The companies will begin offering a joint solution called HP NetServer Assured Availability in July.
Most uptime commitments on Windows NT right now involve the use of two-node Microsoft Cluster Service. The approach requires the Enterprise Edition of Windows NT and special cluster-aware applications.
Marathon’s method owes its heritage to expensive fault-tolerant hardware systems that offers full redundancy for every component. Instead of using a cluster, Marathon ties four standard Windows NT servers together. The solution separates the computing functions from the input/output functions in separate NT servers.
A computing server connects to an I/O server in what Marathon calls a tuple. Then the computing servers are hooked together and perform in lockstep. If one goes down, the other continues the job without the failover delay or application reboot required in clustering situations. Meanwhile, the I/O servers perform asynchronously, with each linked only to its associated computing server. That way the confluence of events that causes one I/O server to crash is not repeated in the other.
"This technology does not try to fix or patch or modify NT in any way," says Jon Affeld, worldwide product manager for HP NetServer Assured Availability Solutions. "Its basic strategy is it insulates or masks the user from failures. You will have blue screens on one of the servers, but the odds of that occurring on the other server at the same time is very small. What it means is that the array as a whole is resilient to those types of failure."
The HP-Marathon systems work with off-the-shelf Windows NT Server, shrink-wrapped applications that do not need to be cluster aware and standard HP NetServer components. Therefore, Marathon’s approach offers fault tolerance at a much lower price than traditional hardware approaches to fault tolerance, plus it also allows users to plug the latest servers and fastest Intel chips into the systems.
The 500 systems Marathon has sold since rolling out the technology in 1997 are a fraction of the number of Microsoft clusters on the market. "I think certainly the HP relationship puts some fairly big muscle behind Marathon and its approach," says George P. Lester, senior analyst at Harvard Research Group, market analysts that focus on high availability. "I think, relative to the market, it’s still going to be a smaller portion. But I think it’s going to be a fairly good size and growing in this area."
HP will sell the solution as an integrated system pre-configured by HP or an authorized reseller. The company, one of the first to come out with a 99.9 percent uptime on NT guarantee, hasn’t defined the packages it will offer through the Marathon partnership.
"We’re evaluating how quickly we can offer uptime commitments," HP’s Affeld says. "There are a number of things you need to do before you can offer five nines [99.999 percent uptime]. The technology is fundamentally capable of that."
A 99.9 percent uptime guarantee allows just under 9 hours of downtime per year. A 99.999 percent guarantee permits only a little more than 5 minutes of downtime and goes beyond the level of most service level agreements in Unix environments.