Some questions abt MogileFS

[James Byers]

> 1. Do you think MogileFS is suitable for the following uses?
> a. Photo storage for online album

Yes.  Members of this list probably have in aggregate hundreds of TB  
of mogilefs storage set up to handle photos and images.

> b. Email storage for email system

Maybe.  If you've built a webmail system for example that needs a  
storage backend, you might use mogile for this purpose.  But mogile  
is not suited for traditional mbox/Maildir/etc. storage.  MogileFS is  
not a posix filesystem, and it isn't intended to be.  It's accessible  
via a simple key->value storage and retrieval API over HTTP (with  
namespaces for replication policies, etc.)  Though some have modified  
mogile to store file paths as keys, it doesn't really do this out of  
the box.  Manage paths in your application, use mogile to store and  
replicate files across as many machines and disks as your application  
requires.

> c. Cache storage (i.e. static HTML publishing, similar the one used in
> MT, directly send the static HTML file to client rather than doing DB
> query each time)

MogileFS is not a cache.  See its cousin, memcached.

> 2. How scalable it is? Any real word examples, e.g. how many servers,
> total storage size, etc.

Tens to hundreds of servers, tens to hundreds of millions of files,  
many TB.  We've got several million small files stored, a few hundred  
GB.

> 3. What are the advantages in using MogileFS over GFS (Red Hat) or
> Coda? (Besides they are difficult to setup).

As I alluded to in the 'email storage system' question, mogile is not  
comparable to gfs or coda.

James
Wikispaces.com