The problem is twofold:

1. Redhat releases about twice as often as the other unix vendors we support. Supporting twice as many OS releases increases DCS' labor costs pretty noticably. For example patch labor for four OS releases is very nearly two times as high as patch labor for two OS releases.
2. Redhat stops issuing new patches for some releases pretty quickly, and it is these releases that we are unwilling to support. More specifically, redhat really only commits to a reasonable duration of patch releases for the last minor revision within a given major release. So 4.2 and 5.2 will be supported with patches for a while after their release (because there was no 4.3 or 5.3), but 5.0 and 5.1 aren't and similarly 6.0 and 6.1 probably won't be. It is unknown if 6.2 will be the last minor release within the 6.x major release. If there's a problem with 5.1 (for example) and redhat chooses again not to issue a patch for it, DCS may end up expending significant effort to create a fix that isn't even integration tested by redhat.