On Wed, Mar 26, 2008 at 11:08 AM, Chris Withers <chris@simplistix.co.uk> wrote:
Tim Nash wrote:
I don't have the skills but I think it would be cool if some student ported Zope to utilize features of the google file system or libferris. Libferris is a virtual file system that mounts just about everything including postgres, xml and OpenOffice docs.
Where are you thinking of plugging Zope into this storage layer?
Something like this: Apache -> zope (business logic, security) -> libferris along with ZODB Perhaps making all libferris resources url addressable within zope. However Postgres and other databases would be better off getting accessed through zsql. Perhaps also make all ZODB objects published as libferris resources.
If zope ran on the gfs (primarily adding business logic, security and publishing) it would give a boost to the value of any zope based company.
I don't know what "running on gfs" would mean, but surely this is only available to Google employees and internal projects?
The google file system is a distributed file system. http://labs.google.com/papers/gfs.html If the zope <--> distributed file system interface was clean, zope could run against other distributed file systems. Then when the gfs becomes a service that can be purchased (surely coming one day) zope is ready to be ported to gfs. Also, google can only offer so much without the ability to utilize business logic and security. They must be looking for an affordable way to offer business logic that can scale and serving secure objects. A bright student can help them explore using zope for this.
Also, I think it would be fun to run map/reduce on my stored objects!
Dunno what this means either...
Here is a paper on map reduce. If zope is storing documents, it would be useful to map / reduce the documents stored in zope and prepare reports, etc.. http://labs.google.com/papers/mapreduce.html Here is an overview offering many competing arguments: "Why Should You Care About MapReduce?" http://www.theserverside.com/news/thread.tss?thread_id=48283 Also, check out hadoop. http://hadoop.apache.org/
Chris
-- Simplistix - Content Management, Zope & Python Consulting - http://www.simplistix.co.uk