But Perforce gives you local files and does not rely on NFS like ClearCase (unless you set it up that way). So, you need to compare ClearCase access times to local access times. Defeinitely do this on an NT machine if you have NT in your mix. And, keep in mind that if you're going over a WAN, ClearCase performance will suffer. Perforce will too but, only slightly in the getting of files and not with compiles. Also, when the Perforce server goes down, you can still use the files you have - with ClearCase, you cannot.

The main thing that I notice though with Perforce is that it really optimizes the command set for batching operations. Example: In Perforce you just say submit and it knows the files that you have checked out. In ClearCase, you have to pick the files that you want to submit (maybe this is fixed in their UCM feature) and directories are often forgotten about leading to nightly build breaks. This UI design carries through to other operations like merging a set of files or listing contents of a label, etc. A lot of ClearCase operations involve first creating a view and choosing the appropriate config spec before running query wheras Perforce allows you to do a lot more by simply running the client command.

I was the Piccolo (predacessor to Perforce) administrator at Ingres for 2 years before going to Informix in 1994 so I speak from direct Perforce/ClearCase experience with very similar environments - big database company, multiple UNIX platforms, NT, multiple development sites, hundreds of developers, etc. ClearCase is definitely a more costly product to purchase, administer and use. While it is very good in many ways, I find that the basic concept of the MVFS filesystem incurs more of a performance penalty than the file control benefits warrant.

If your seriously considering ClearCase, do yourself a favor and really think about the monetary costs, performance and availability implications of placing the central repository in every file access operation. Dig into their new UCM and snapshot views functionality and consider whether this is a reasonable solution (and that it looks a lot like Perforce anyway) and, again, whether it's really worth 6 times the money (not to mention the additional server, network and software (NFS Maestro for NT clients, ClearCase MultiSite for every user even though your just going to have a few sites...) I just went through the same kind of evaluation and couldn't get behind ClearCase. Bigger is not always better.

Maurice