You essentially need a fast network card and a slow disk subsystem to ever see this problem but it is definitely worth noting.
http://support.microsoft.com/kb/907404
This explains it better than I can:
This problem occurs because the available memory has been consumed. The available memory can become consumed when the disk subsystem throughput is slower than the network interface. The cache manager should buffer the difference in throughput between the disk subsystem and the network interface. However, hotfix 839406 and security update 896422 bypass the mechanism to throttle back these incoming writes after the data exceeds a certain threshold. If large writes are sustained for a long time, other critical writes will be deferred for a long time. When this occurs, the computer may stop responding.