Warren Togami wrote:
[EMAIL PROTECTED] wrote:
snip
Perhaps my understanding of these tools is inaccurate, but I thought that nothing Open Source came close to the capabilities of tools like Ghost. For example, we have nothing like the multicasting features. None of our tools are nearly as polished either.
Yes, the lack of multicasting is a problem. In theory if you have a "reliable network" (all packets are guaranteed to be delivered, in the order they were sent, without corruption), you can just spew UDP packets to a site local multicast address and have all you comps receive it. Unfortunately Ethernet isn't reliable. Ghost gets around this by having every machine ack every packet as it goes, which is slow, but very reliable.
I have thought of ways to do this. Spew UDP packets to the network via multicast, but with some form of sequencing info. Then when the session is over, each machine requests the data it missed via unicast using TCP. A checksum (MD5 or simlar) can be performed on the final image to make sure the image is uncorrupted. Checksumming of each packet would also be a "good thing" so that corruption could be detected on the fly and replaced during the end unicast session, lessening the chance of undetected corruption requiring a complete retransmit of the image.
The lack of support for NTFS is not a problem in and of itself, but it leads to problems. It's not that we can't support NTFS (we don't need to understand the filesystem at all in order to do a block-for-block copy), but that we can't modify Windows' "magic numbers" like ghost can. Of course, doing so would requrie stable NTFS write support, which we lack. I don't forsee a Norton ghost replacement in the near future in this regard.
As for polished, this is just because nobody has bothered to do so. Writing a simple curses based gui (heck, you could even use dialog!) would be quite possible.
--MonMotha
pgp4qTPyRmJuc.pgp
Description: PGP signature
