Bandwidth hunger creates a Cache-22
If the folks at Cache Now! and the National Laboratory for Applied Network Research (NLANR) have their way, copies of your Web site will be pasted on dozens of servers in the U.S. and around the world without your knowledge or your permission.
The problem is bandwidth.
Your personal client browser, Netscape Navigator or Internet Explorer 4.X, caches recent Web site visits to your hard disk. The next time you visit the same site, the browser first looks to see if a copy of the requested URL is already in your hard disk cache. If it is, the load time is remarkably faster than the usual dial-up speed. Thus, personal caching is good.
At the backbone level, lack of sufficient bandwidth is a global problem. Internet telephony, cable modems that suck down data in the last mile at rates of 400K to 1.5M bit/sec, videoconferencing and multi-media requirements are filling the pipes of cyberspace as fast as fiber is installed. Thus the understandable desire to find a way to offload some of the massive bandwidth demand.
The proposed solution is simple. In addition to your personal cache, another cache would be added at the network level. Your Internet service provider would maintain an active cache of the most-often surfed Web sites and when you request that URL, instead of being routed to the actual server, your download occurs from the local ISP-maintained cache server. The result is a faster download speed.
Cache Now!, NLANR and a number of supporting international organizations want to go even further. They are promoting a hierarchical cache structure.
Let's say you request a URL from England. First, your browser would examine the client cache. If the URL is not there, it would then examine the network cache, then the regional U.S. cache, then the national U.S. cache, then the national U.K. cache, then the regional U.K. cache, then the network ISP cache. If the requested URL is not found at any of these points, then and only then will the browser access the actual server hosting the URL itself.
On the surface this method sounds sort of neat. RAM and CPU time are less expensive than bandwidth and require minimal infrastructure construction. Also, network bottlenecks might become less severe and network speed would increase.
But what about the effects on privacy, security and control over Web-based content? Some of my concerns include:
As the Web becomes more commercial in nature and as more and more private information is stored and moved across international boundaries with little constraint, the problems associated with national and regional caching become more obvious.
It seems clear that if the problem is a lack of bandwidth, then the answer is to add more bandwidth, not a Band-Aid that's not big enough to cover the wound.
For a complete look at both sides of network caching, check out www.goforit.com/cache and ircache.nlanr.net/Cache.
