The difference between grid computing and cloud computing

Lew Tucker, Cisco Systems' CIO for cloud, is no stranger to large-scale computing. Tucker, who joined Cisco in June, previously worked at Sun Microsystems, where he was responsible for SUN's cloud and large-scale network computing.

Tucker's main responsibilities at Cisco will cover different business units, because the cloud will involve enterprise users, service providers and partners. For Tucker, the cloud is not just grid computing or other new names for large-scale distributed computing, although they have commonalities.

"The cloud is very relevant to a lot of what we know from grid computing," Tucker said in an interview. "We know the benefits of scale and having a unified infrastructure, and we understand that you have to automate the management."

Grid computing was the most common term between 2000 and 2005, and SUN has played an important role in popularizing grid computing. In 2004, SUN launched an enterprise-level grid platform to provide utility computing for enterprise users. Although the early promises of grid computing to users sounded the same as the promises of the cloud, Tucker emphasized that there are some obvious differences between them.

"The big difference with Sun's grid computing is that it was designed to run very large-scale applications in batch mode," Tucker said. "So grid computing is very suitable for very large-scale computing, where applications send data to the grid for calculation and then output."

In contrast, the cloud is not a batch processing system. Unlike data centers and server platforms, the cloud is based on the network application software delivery space. Tucker emphasized that the concept of the cloud originated from large cloud companies such as Amazon, which must build large-scale infrastructure to run one or two applications.

"With the cloud, we can try to accomplish the same things on a large grid-like infrastructure," Tucker said, "but currently that infrastructure is designed to serve long-running applications that are delivered over the Internet."

From a management perspective, the broad nature of the cloud makes it difficult for companies like Cisco to actually measure how much of their business is cloud-related.

Cisco has a unified computing platform which is a cloud platform, but they also have a variety of switch products.

"We first have to really understand what is cloud and what is just traditional network business," Tucker said.

Tucker also stressed that analysts have different forecasts for the cloud market, and he is also collecting more detailed statistics on the current state and future of the cloud market. His plan is to measure Cisco's capabilities in this regard by looking at the market as a whole.

"The Unified Computing System is an important product that users use in the cloud, but most clouds today may run Cisco network equipment," Tucker said. "They may not be able to use the Unified Computing System at this time, so we try to build the cloud with Cisco network equipment, which is not a Unified Computing System and is generally considered to be a traditional infrastructure."

Looking ahead, Tucker laid out several goals for Cisco's cloud business. At the top of the list is to send a clear message to the market about what Cisco's cloud strategy is so that all the different cloud divisions within Cisco can work together as a whole.

"Cloud computing is very important to Cisco, and we have to find ways to expand our efforts," Tucker said.

Tucker also emphasized that Cisco's sales are to both enterprise users and service providers, so they can also help enterprise users use private clouds and ultimately benefit from service providers.

"What I want to say is that enterprise users can move freely between private clouds. We can help them build private clouds, and we can also help them extend their workloads to the enterprise-class cloud computing provided by Cisco, thereby benefiting from service providers."