Compressed caching mostly benefits from the increases of CPU speed relative to disk latency. Nevertheless, a different factor comes into play when disk and memory bandwidths are taken into account. A first observation is that moving data from memory takes at most one-third of the execution time of our WKdm compression algorithm. (This ratio is true for both the Pentium Pro 180 MHz machine, which has a slow memory subsystem, and the SPARC 300 MHz, which has a fast processor. It is significantly better for the SPARC 168 MHz machine.) Hence, memory bandwidth does not seem to be the limiting factor for the near future. Even more importantly, faster memory architectures (e.g., RAMBUS) will soon become widespread and compression algorithms can fully benefit as they only need to read contiguous data. The overall trend is also favorable. Memory bandwidths have historically grown at 40%, while disk bandwidths and latencies have only grown at rates around 20%. (An analysis of technology trends can be found in M. Dahlin's ``Technology Trends'' Web Page at https://www.cs.utexas.edu/users/dahlin/techTrends/ .)