I recently had the opportunity to talk with Steve Keilen and Michael Bilancieri from Marathon Technologies about their high availability products for XenServer. Marathon builds on to their existing HA products by releasing version 5 of everRun for XenServer.

I think that is an important point when it comes to product maturity. According to Steve, “everRun 2G stands for the second generation of our traditional everRun products. These are the products that have been very successful in vertical industries such as process automation and emergency dispatch where Windows has infiltrated and become business critical. What we are doing is bringing a load of features and functions to that traditional market. everRun 2G also adds in more supported platforms including Windows 2008 and 64-bit versions as well as allows for multiple applications to be protected on a single OS instead of just one.” Michael pointed out, “There really is no grey area for mission critical systems requiring fault tolerance. Our offering is version 5 that has been in production for years protecting mission critical applications such as emergency dispatch systems. What the competition is offering is a version 1 beta product. We’ve been working on this product for 15 years and solidified and hardened it, so it’s extremely stable and extremely robust.”

I remember teaching XenServer 3.2 class in Boston a few years ago and that is where I was introduced to Marathon Technologies. They had a couple of their people attending the class and we talked about their HA capabilities. I’ve been impressed by the stability and ease of use of the everRun product.

The second announcement is for the everRun VM Lockstep for Citrix XenServer. Michael said, “Ever since we partnered with Citrix and integrated our HA capability into XenServer, people have been asking us when we were going to add in our everRun FT capabilities as well because we still have the zero downtime requirements but would like to consolidate the server workloads on XenServer. This is the first software based system level fault tolerance for virtual environments. This is true system fault tolerance. We lock-step the virtual machines between one or multiple XenServer hosts. So if you lose a device, a network or storage component or the entire host, the application will continue to run with no interruptions, no loss of data or transactions.”

They have it set up where you can set the level of protection per virtual machine in XenServer. “We refer to this as Level 3 (where HA that comes with XenServer is Level 1). Now customers can deploy their most mission critical applications into their XenServer pools and are not limited to just two hosts. We support multiple servers at the same time running full Level 3 fault tolerance. This supports the free version as well as the essentials. It’s important to point out that VM Lockstep protects the entire stack, from the application, operating system, server hardware, network and storage. The standard approach is to try to create high availability from each of the components such as storage, network and the servers, and then get them all to work together. That’s a lot of work. If any one piece of that chain doesn’t work, your fault tolerant solution is no longer fault tolerant. We harden the whole system, from application, operating systems, storage, network and server. If something fails, the application keeps running.”

The architecture looks like the following graphic. everRun is installed and runs on XenServer under the virtual machines. “We install right on XenServer and it sits there pretty benign.” according to Michael.

I fall into the same redundant device thinking just as everyone else does that architects projects. The one point that jumped out to me was one Michael made, “You don’t need shared storage because we have it covered. You can deploy a completely fault tolerant system at a different geographic Disaster Recovery (DR) site on one server with local storage. If the primary site goes down, the application still runs with no data loss at the DR site. Many customers choose use because they don’t want to buy another SAN for their DR site just to protect a few mission critical systems.”

I think this is really a compelling product announcement that should be considered when designing for zero downtime fault tolerance systems for Windows. I’d like to thank Michael and Steven for their time.