Glassware clustering on the other hand is intelligently designed with the application being requested in mind. If a customer has deployed Glassware appliances at several regional offices as well as at a central data center, the customer can distribute the hardware. The load balancing algorithm is quite simple (users accessing graphics apps go to the local Glassware server first; lighter apps go to the data center). No server is required for load balancing because Glassware knows which app the user is going to use and Glassware knows the typical resources requirements for each app. We call this ability to connect the user to the right application on the appropriate appliance Context Enabled Switching.
Each installation of Glassware can serve as a separate backup for the others. For BIG applications, because everything is done locally over the LAN (the compute and the storage), performance is fantastic. The real time back up of files may lag as the file moves from the regional office to a central backup location but that is unseen by the user. In a worst case storage scenario, the user would lose seconds of changes at most. If a Glassware server fails, the user automatically fails over to another location. Since the Glassware servers are so efficient the Glassware servers at other locations will have no problem handling the extra load. Glassware’s Clustering technology enables the customer to publish applications across a geographically distributed environment of appliances that can both scale up (more servers in the same location) and out (more servers but at other locations) …almost infinitely; or as we like to say “with Google scale”. Glassware Clustering also creates the flexibility for organizations to put appliances closest to their users to eliminate network latency while enabling real-time failover, and providing performance certainty.