Marathon Technologies has been working on creating a fault tolerant environment based upon standard, off-the-shelf systems for quite some time. everRun MX is the newest in a long line of products designed to bring the benefits of fault tolerant computing to more organizations by using industry standard servers. What’s new this time is that the product does its magic using symmetric multiprocessing (SMP) as well as multi-core systems. Applications running in a well-implemented environment should continue to run even though some components have failed.
Here’s what Marathon has to say about this new release
Key attributes of everRun MX
- World class application availability through downtime prevention instead of recovery.
- Full SMP/multi-core fault tolerant and redundant operation across servers eliminates risks associated with system/VM restarts and data recovery.
- No hardware lock-in. Runs on commodity servers, not specialized equipment. It immediately leverages hardware innovations while providing server selection freedom.
- Natively scales and works with all applications, regardless of computing demands.
- Requires no application level customization or complex scripting routinely found in conventional availability offerings.
- Fully automated and self-healing. everRun MX’s prevention response requires no operator intervention, a trademark Marathon capability ideal for environments where staffing or access is constrained.
Snapshot analysis
As Professor Marcello Truzzi once said, “Extraordinary claims require extraordinary proof.” Another often used quote is “The proof of the pudding is in the eating” which is attributed to William Camden. Claiming the ability to create a never-fail environment based upon standard off-the-shelf hardware is something that most would say is an extraordinary claim. After all, systems, memory, storage, networks and all other components of a computing solution fail at one time or another.
Hardware oriented fault tolerant systems suppliers, such as Stratus, would tell you that installing a machine that has built-in redundancy and hardware failover mechanisms would be less costly than building out an environment using multiple independent computing systems. After all, they would say, a single copy of the operating system, data management tools, application frameworks and applications themselves would be needed. That alone would reduce the software and administrative costs of a solution.
Marathon, on the other hand, believes that a properly designed approach that is based upon common, everyday systems can approach the availability and reliability of those purpose-built systems. I suspect, however, that they can’t approach the failover speed seen when using those systems.
When asked for proof, Marathon would point out that their technology is in day-to-day use in 30 countries by thousands of organizations. Is that enough proof?
Well, it certainly seems convincing. Has your organization used one of Marathon Technologies’ everRun products? Has it lived up to their promises?
If your organization has a need for this type of solution, it would be worth taking the time to see a demonstration of this technology.
