[Gluster-users] BDB speed benefits

[Anand Babu Periasamy]

Hi Stas,
BerkleyDB storage backend makes GluserFS a distributed database file system, efficient for 
very small files (less than 10KB).

It also allows you to read and write to a file with out even opening it, using standard 
extended attribute system calls (in a single atomic transaction). "Extended attribute get" 
with key as file name will return the entire content of the file. Similarly it also works 
for "extended attribute put" with key=FILENAME and value=CONTENT to create and write to
the file.

Files will be stored as BDB records. Each directory has its own BDB file. If you have
a cluster of 10 nodes, each directory will be backed by a cluster of 10 BerkeleyDB
files. Every new directory you create, a new set of BDB files will be alloted across
the cluster.

It is a recent addition to GlusterFS 2.0.x codebase. Do *not* use it for production,
until it is widely tested by the community. Please let us know if you come across bugs.

Happy Hacking,
--
Anand Babu Periasamy
GPG Key ID: 0x62E15A31
Blog [http://ab.multics.org]
GlusterFS [http://www.gluster.org]
The GNU Operating System [http://www.gnu.org]



Stas Oskin wrote:
> Hi.
> 
> Any idea what speed benefits the BDB translator provides over standard 
> file storage?
> 
> Also, how it's reliable, and what's the maximum file size it stores in 
> the DB?
> 
> Thanks.