Marathon everRun MX review

Today's servers incorporate many fault tolerant features including RAID arrays, multiple network ports and hot-swap power supplies but the processors, memory and even the systems themselves still represent single points of failure. Businesses that can’t afford even a second of downtime for critical applications need total redundancy across the board.

Marathon Technologies specialises in fault tolerant systems and claims its latest everRun MX is the world’s first software product to provide full protection for SMP and multi-core aware Windows applications. In this exclusive hands-on review we put everRun MX through its paces in the lab and see how well it handles system failures.

Marathon doesn’t lock you in with proprietary hardware as everRun MX supports industry standard servers. The two servers don’t need to be perfect twins but they should be as close a match as possible. However, the new FT-SMP feature, which supports SMP and multi-core applications, requires that the processors on each server be identical and they must be either 5500 or 5600 Xeons.

It’s worth taking a closer look at what lies beneath everRun MX as it is completely different to the HA version which we've already reviewed. First up is the base OS, as the servers run Citrix XenServer with everRun MX installed on the hypervisor.

The biggest difference is that HA only supports one VI (virtual instance) so it can only ever provide a single fault tolerant virtual server. Furthermore, no matter how many physical cores you have, HA can only present one core to the VI.

MX allows you to create as many VIs as you want and each one can have up to eight cores. The only limit is the load placed on them and what the hardware can reasonably support. Unlike HA, which can only run Windows Server 2003 32-bit, it also supports Windows Server 2008 or 2008 R2 and all 32- and 64-bit versions.