Jack Ungerleider wrote:
Some of the dsicussions in the Client Side Zope thread got me to thinking about an article I read in Linux Journal sometime ago on the Coda file system. As luck would have I can't seem to find that issue. Most of the information contained in that article (now over a year old) can be found in updated form at http://www.coda.cs.cmu.edu/ .
For those who've not heard of coda its a distributed filesystem that supports a variety of the features, including disconnected operation and replication among multiple servers, that were mentioned in the previous thread.
I will admit to knowing only the very surface information about coda and throw this to the list for anyone who has actually used it. Also do any of the DC folks think that a data store centered around a coda type fs would be doable?
Absolutly, but in this case you can just use FileStorage, since coda is a filesystem.
I've been intrigued by coda since I first read the article this may be the reason to explore further.
Any other thoughts on this?
I attended the Coda session at last year's Linux Expo. There, the professor in charge of the project gave a very clear explanation of coda and how it works. Useful results with Zope should be immediate, Simply store your Data.fs file on a filesystem that is replicated by Coda. There is a more interesting application however, and that is implementing the actual coda protocols themsleves into a sort of ReplicatedStorage (I'm not positive if that is the best way to slice the pie, but it's an idea). Another idea is to roll a similar or identical protocol into ZEO, so that not only are Storage Clients and Storage Server are distributed in a parent/child relationship, but also that Storage Servers can be distributed in a peer relationship that replicate each other to provide failover and replication. Interesting stuff in any case. I'd really like to hear about someone who has a success story with Code and Zope using just it's filesystem capabilities. -Michel