Pre-sales questions

[guest]

I have been evaluating the Pro version of Vice Versa this week. It appears to be almost ideal for our needs. We have a set of data -- ~200K files -- that needs to be in sync across multiple systems. Only a relative handful of these files change on a daily basis, but keeping track of what needs updating has proved to be a major pain. Enter Vice Versa.

The first question regards the synchronization methods. The bidirectional sync + history looks good, with one exception. This appears to be the only method that allows both full updating and deletion of unwanted files. The problem is that we do not want *everyone* to be able to perfom deletions. The prospect of any one of a dozen folks deleting a file accidentally and thus killing it from all locations is not a good one.

We are running with a centralized server. All satellite machines add new and updated files to the central server, and download new files + updates from the server. Is there a way to designate only a few systems as having delete power over the server files? In other words, all systems would add new and updated files to the server, download updates, and delete local files that were deleted from the server. Only a selected few systems would be able to delete server files, and thus propogate these deletions to all other computers.

Question two: We operate in enough different time zones and countries that keeping track of time zone offsets is not feasible for a synchronization system that does not understand the difference between local and remote times (hint... SFTP would be *really* nice!). I have tried using both VPN and WebDAV over SSL. Neither allowed having easy sync based on file times -- too many variables of countries that have daylight savings, do not, etc. The CRC check works reasonably well, but by chance, we found a pair of files that generated a CRC collision. These were two versions of the same binary file that VV evidently calculated identical CRC values for. Most of our files will, when updated, have differences in the first 10-20K of data. I had VV set to calculate CRC values based on the initial 25K to save bandwidth (not wanting to download 30GB+ each day for synchronization). Would it be possible to implement a less collision prone algorithm to detect file differences? MD5 is not perfect in this regard, but it is pretty good. Tiger might be better, and should be as fast or faster. Even an SHA-x hash would probably suffice, and would require less computation.

Question three: Could full WebDAV support be added? This looks to have faster performance over a secure connection than most VPN setups. As it stands, however, Vice Versa can not directly connect to a WebDAV share. Instead, it needs to go through a dirve emulation program which distinctly slows down all directory traversals and file lookups. Alternatively, SFTP could be used, but this is falling by the wayside.

[TGRMN Software]

1- You could use backup (mirror) from the server to the client for those clients that can only receive data but not change or delete files on the server.

2- CRC and SHA only tell if files are different but not which one is the newer / older file. So these hash calculations are not good for sync. They are good for file verification, when you want to to make sure that source and target are identical. I am afraid they would not be helpful in your case....

3- We are considering WebDav, FTP and SFTP for future releases. FTP is a weak protocol, however, for file sync, because file timestamps are not persistent (they get changed when you upload a file).
_________________
--
TGRMN Software Support
http://www.tgrmn.com
http://www.compareandmerge.com