Not Your Father’s Cache

I’m dating myself here, but on this week back in 1997 I found myself at the Las Vegas Interop show (then called Networld+Interop) announcing a new product called Shared LAN Cache Server (aka SLC Server) for a then “new player” in the network acceleration market called Measurement Techniques. We promoted SLC Server as “a lower cost alternative to deploying additional file servers at each remote office” that “[reduced] file access times and [eliminated] redundant read requests for users in remote offices that access files over a WAN.”

Essentially, SLC Server cached all non-rights-related files – application software, images, electronic messages and other data files – on the local client, thereby off-loading the WAN.

Although caching emerged as a potential way of accelerating the performance of specific applications, such as web applications, while reducing overall WAN traffic, it found only limited success and the market for such products ultimately subsided as the result of several operational and functional limitations.

Nevertheless, even today, caching can provide a reasonable boost in performance. Some like Dino Esposito at the independent Microsoft resources blog, DevProConnections see caching as a key to scalability, while George Crump at InformationWeek says, “Caches are ideal for situations where different applications could use a performance boost at different times.”

Unfortunately, there are still significant limitations when using a cache across a distributed enterprise. Some of the problems are that caches:

  • Are application specific
  • Have version synch issues
  • Require exact matches
  • Demand significant IT support and management
  • Point to a cache proxy server instead of to the original application server
  • Handle only static content

For smaller companies with fewer clients and very little dynamic data, caching can be a reasonable solution to off-loading the demands of a WAN. However, in larger enterprises, there needs to be a more comprehensive WAN Optimization solution.

Caching In

On the surface, the Silver Peak solution sounds similar to a web cache in as much as it optimizes WAN performance and reduces traffic. However, the Silver Peak solution addresses many of the fundamental deficiencies associated with web caching, including:

  • Application Transparency: Silver Peak works at the network-layer of the ISO stack and therefore is able to provide performance improvements across all enterprise applications, regardless of the transport mechanism (TCP, UDP, etc).
  • Matching traffic patterns: Network Memory can detect when the same information is sent using different applications, and detect when modifications are made to existing data.  This enables Silver Peak to better utilize WAN bandwidth and provide better application performance than caches
  • Support for dynamic content: Web caches are dependent on having cache tags set properly on web objects.  Silver Peak does not require any pre-defined tags to detect repetitive data.
  • Seamless Integration: All requests for information from a server are delivered to the application server itself, not a proxy server, and the server’s exact response, not an old cached version, is delivered to the user
  • Data coherency: All application locking semantics and file/record locking capabilities are still performed by the native server, not a proxy device, ensuring 100% data coherency.
  • Security, compliance, and management:  Access control policy mechanisms are centrally maintained within the servers themselves, eliminating potential security risks and avoiding unnecessary management headaches that come with replicating and maintaining access privileges across multiple devices

Silver Peak’s Network Memory provides improved performance over caching, without the inherent management and operational limitations. By delivering a network-layer solution that leverages advanced byte-stream fingerprinting technology and high capacity local data stores, Network Memory provides order-of-magnitude performance gains across almost any enterprise application.

In addition, Silver Peak localizes information, while centralizing the control of applications servers and storage, making Network Memory a much more indispensable tool than caching for enterprises looking to “cash in” on the cost and management savings associated with WAN Optimization and data center consolidation.

About the author
Jonathan Bloom