The latest version of Marathon's everRun family, everRun MX, is targeted at small- to medium-sized IT operations as well healthcare, manufacturing and remote offices. everRun MX synchronizes between disk systems, and can work with storage area networks (SANs) or direct-attached storage (DAS). Marathon's first generation everRun 2G was introduced about two years ago and it supported a single processor. This generation of everRun MX supports two, four or eight physical processors.
The everRun MX software is composed of three parts. The Component Manager allows the application to look at the hardware across all platforms as a single instance. It controls all the I/O operations to the disk, network and memory and helps maintain connections for data and transactional integrity. The Workload Manager supports multiple workloads so that applications can be stacked onto a single server pair, and the Availability Manager helps the traffic between the two servers stay in sync. "All of our key intellectual property is inside the Availability Manager," Marathon CEO Jim Welch said. "It's where all the hard work is done. This is where all the true fault tolerance occurs."
Pricing for the software is about $10,000, which includes protection and support for two servers.
Marathon vice president of marketing Rob Ciampa said everRun MX is more about preventing downtime than enabling application recovery.
"It's much better never to go down than to deal with the implications of an outage," he said.
Laura DiDio, principal analyst at ITIC, described this launch as a reboot of the company. Marathon secured $13.5 million in venture funding over the past year, and she said most of the vendor's employees have joined in that period.
"This is getting back on track with new staff and new direction," DiDio said. "They have expanded their focus. They are heading more into applications, virtualization and fault tolerance."
She said the most interesting thing about everRun MX is that is that it enables software-based fault tolerant without requiring a specialized hardware solution. "In the past, if you wanted fault tolerance, you needed specialized fault tolerance hardware," DiDio said. "Most people couldn't afford it."