Technology

Cache Differencing

Caching brings information closer to the end user by storing recently accessed data in local memory or on hard disk, and reduces the time it takes to bring back needed information, making the user experience more positive and action oriented.  While today's browsers maintain their own cache, they tend to be overly conservative.  This means they will err on the side of requesting a new piece of data or object, usually when it really hasn't been changed.  This not only impacts response time to the end user, but also saturates bandwidth with unnecessary data transmission.

Cache Differencing takes the concept one step further and maintains identical copies of the browser's cache at the local device and at the Stampede appliance. Our technology then uses intelligent differencing technology to understand what data has actually changed, and then transfers only the changed data. The local device functions normally, but with less data being transferred, you realize improved utilization of the network, and increased end user productivity.

Traditionally, pages can be marked as cacheable and will have expiration dates. When they expire they must be retrieved from the original server, resulting in additional traffic and data being transmitted across the network. The Stampede Application Acceleration Series client caches all pages returned to the browser (even pages that are marked as non-cacheable) and performs validation when needed to ensure that no stale data is returned to the browser. When the browser asks for a page or an item that has expired or been marked as non-cacheable, the client sends a validation request to the Stampede appliance. If the appliance is aware of the last page the client cache contains and can compute differences in the page, it sends just the differences to an expired page or non-cached page. (If the differences are too big or if the appliance no longer has retained the last version that the client has, then the entire page is returned and subsequently cached for future possible differencing). The client in turn reconstructs the requested page, caches it, and returns it to the browser. Checksums are calculated by the Stampede appliance and verified at the client so that pages will never be delivered incorrectly.  While this technique adds value on expired pages, it is extremely effective for dynamic page generation.