While many commercial users of Grid computing use the technology primarily to tackle low-hanging problems like hardware utilization or providing CPU power to data-intensive applications, companies are increasingly realizing that Grid computing can be used to improve the performance of less data-intensive, but equally business-critical applications within the datacenter.
One such approach to applying Grid technologies to transactional applications is called application virtualization, and DataSynapse has been focusing on this idea since the release of it FabricServer solution last October. In fact, according to CTO Jamie Bernardin, the company is more comfortable describing what it does — especially with FabricServer — as “application virtualization,” not “Grid computing.”
The reason, he said, is that application virtualization seems to better resonate with what the company is doing and what they are going after in the market. In addition, Bernardin said, the concept of application virtualization has evolved beyond the typical Grid positioning, as “Grid is many things.” However, he acknowledges, there is definitely a Grid aspect to what DataSynapse is doing.
“We are hosting and managing applications, and doing things like dynamic provisioning, on a group of shared resources that we would refer to as Grid resources,” Bernardin said. “What we're doing is virtualizing the application so it can run on a group of shared machines.”
For DataSynapse, the move toward application virtualization (and the development of FabricServer) stemmed from the success of its GridServer solution. According to Bernardin, customers were coming back to them pleased with the results they were seeing, but also noting that expanding GridServer's capabilities to other applications that make a large footprint in the datacenter would be even more beneficial. Hence, with the release of FabricServer, DataSynapse focused on bringing more application platforms (e.g., JBoss, WebSphere (IBM), etc.) and applications (e.g., Web, BI, BPM and portal applications) into the mix.
“It's not around creating new Grid services,” Bernardin said, “it's really just about putting existing applications onto a managed, shared infrastructure.”
However, DataSynapse isn't the only vendor looking to address the needs of broader application virtualization within the datacenter. According the United Devices CTO Jikku Venkat, UD has been working with customers, including British Telecom, to develop a solution to virtualize business-critical applications. Venkat said UD not only has been seeing demand for such a product from looking to better manage their datacenters, but also from big telcos who want to provide managed services to their customers.
He added that this really will be a new product and solution capability for UD, and that the company is working to enhance its current solution set — including Grid MP — to support transactional applications. The technology, he said, will be packaged a solution for datacenter and mission-critical applications.
Of course, enterprises aren't expressing interest in such a solution out of the blue — they are seeking to take their currents efforts to the next level. “We are being very practical about the way we approach this whole space,” said Venkat. “We understand and realize that a lot of our customers are already engaged in some sort of datacenter consolidation/server consolidation efforts, and we are showing them how to take that next step and apply Grid technology right on top of what they are doing and take advantage of even more value and even more benefits.”
The “next step,” according to Venkat, is to move applications onto a virtual infrastructure based on application demand and application service-level agreements. That's really what application virtualization is about, he said. You are able to encapsulate the application so it can run dynamically where capacity is available and be orchestrated based on the needs and demands of the application.
Speaking specifically about how application virtualization works with application server environments, Bernardin said it is about “bringing up more instances of a WebLogic cluster or WebSphere cluster and controlling — based on schedule or priority — what [applications] are running at what particular time.”
Although application virtualization is described by both Venkat and Bernardin as a different approach to utilizing Grid technologies, both are careful to point out that it is simply an evolution — far from the infrastructure overhaul many enterprises fear when thinking of Grid computing. “It's not really as much of a disruptive technology. In fact, it's the opposite,” said Venkat. “It's completely non-disruptive to a roadmap they already are on that we are simply plugging into.”
Reiterating application virtualization's close relationship to traditional Grid computing, Bernardin said: “You could describe shared resources as Grid resources, and whole initiative as Grid computing, except that Grid computing means a lot of things to a lot of different people and doesn't necessarily hit virtualization meaning right on the button.”