Version 1.1
User's Guide

RAM Cache: Accelerating Web Access Previous | Next
Contents

WebDoubler includes two cache managers that you can choose from for your network. The first option, the RAM Cache, holds the entire contents of the cache in RAM. This cache offers very fast throughput but provides a relatively small cache space. The second cache, the disk-based Cache Manager, can maintain much larger cache stores, but adds additional overhead to server operation. The RAM cache is simpler to use, requiring no setup at all, and provides excellent "out of the box" performance. For this reason, it is installed as the default cache plug-in and is covered in this chapter. For complete information on the Cache Manager, see chapter 13.

Important Note! Do not use the RAM Cache and the disk-based Cache Manager at the same time. Only one of these plug-ins should be placed in the active plug-ins folder at any one time.

Theory Of Operation

In all caches, including the disk and RAM caches built in to every Macintosh, cache performance does not improve with size in a linear fashion. For example, a Level 2 RAM cache is included in most modern CPUs that may range from 512 KB to 1 or 2 MB. Increasing the cache from 512 KB to 1 MB (1024 KB) does not, however, double the cache's performance.

What is important is the cache hit rate, or the number of data requests that can be served from cache. In general, as the cache size grows, so does the cache hit rate. However, this improvement is a case of diminishing returns. Cache hit rate improves less and less as the size of the cache space grows.

For a Web cache, this effect will vary depending on your organization's Web usage patterns. If users on your LAN have very diverse Web surfing needs, then cache size may have a more noticeable impact on cache hit rate. For networks where users frequently access the same content repeatedly from many workstations, a smaller cache can be nearly as effective as a larger one.

The RAM Cache permits only relatively small cache sizes, but performs its task very quickly. Since a cache must be maintained for every request, including requests not served from cache, this performance improvement can be seen on every file that passes through WebDoubler. The idea behind the RAM Cache is that this performance improvement will, in at least some cases, more than offset the drop in cache hit rate.

Who Should Use The RAM Cache

We recommend that you try both caches and use the one that subjectively "feels" better to you and your users. It is important to give each cache a few days or a week or so to "burn in", and you should test the performance under the load of your entire network. The RAM Cache will probably "feel" faster initially, due to it's quicker processing. After sustained usage (with the larger cache filling with more data to be served at LAN speed), the disk-based Cache Manager may provide better overall performance.

The more RAM you allocate to WebDoubler, and the less varied your organization's Web surfing, the greater the chance that the RAM Cache will provide maximum performance. If your proxy is allocated less than 32 MB or so, and if your LAN users tend to surf very diverse sites, then the Cache Manager will almost certainly provide superior performance.

Using The RAM Cache

The RAM Cache is installed and ready to be used by default. However, if you have switched to using the WebDoubler disk-based cache, you will need to reload the RAM Cache. First, shut down WebDoubler. Move the cache plug-in, "WDPI Cache", from the "Plug-Ins" folder to the "Disabled Plug-Ins" folder. Then put the "WDPI RAM Cache" plug-in into the "Plug-Ins" folder and restart.

There is no configuration or setup necessary to run the RAM Cache. In fact, no configuration options are offered at all. To disable the RAM Cache, move it out of the "Plug-Ins" folder. The RAM Cache will automatically use the RAM available to it, so to increase the size of your cache, increase the memory allocation given to WebDoubler. This is done in the Finder using the "Get Info" dialog box.

The RAM Cache will store the cache contents to disk when WebDoubler is shut down, if space is available. The contents are then loaded at startup, to preserve the cache contents when WebDoubler is restarted. Note that this will cause WebDoubler to pause when shutting down and starting up, depending on the size of your cache and the speed of your hard drive. For caches of only 10 MB or less, the pause will be only a few seconds. Larger caches, however, may take a considerable amount of time to be saved and reloaded. Note also that the cache contents will not be saved when you change WebDoubler's memory partition, as described in the "Installing WebDoubler And Basic Configuration" chapter of this User's Guide.


Copyright © 1999 Maxum Development Corporation
http://www.maxum.com/
Previous | Next
Contents