I guess this is a popular idea. :-)
I've built what I call an "object -> cloud" mapping that works with
Hbase. It stores everything as json so if you manage to build a
binding for it (through Hbase's thrift interface) it should work with
any language. I say "object -> cloud" because the idea of the
software is to expose to other people some of the features and
limitations of a column oriented database, not just have them use the
same old relational models and end up getting nasty surprises later.
It was designed for read-heavy, interactive applications and
prototyping. For the latter, it has a simple "disk file" version so
people can get the app working before sending it to an hbase cluster.
It allows you to have multiple searchable "columns" by building it's
own indexes off to the side. It tries, for example, to expose the
idea that this is not cheap by making people select the searchable
items in their data.
If folks are interested, please drop me a note. I need to make a case
that this software should be open sourced and several emails from
people about how they might use it would help.
thanks
ian
ps. If anybody has an hbase cluster that has some idle time I'd love
to get some performance numbers... my current unemployment makes
renting EC2 nodes a bit out of bounds...
On Oct 27, 2008, at 2:12 PM, arungabriel wrote:
Beautiful that you have thought of doing an ORM for HBase.
We are planning to store Serializable objects (which holds state)
in HBase
as byte array. We would like to use the information for scheduling
purposes.
Is this really possible practically? Or can i get the String values
in the
object and persist using the ORM?
It would be great if you could share the ORM with us. Let me share
with you
what i came up with :-)
--
View this message in context:
http://www.nabble.com/Hbase-ORM-any-one-interested--tp19739869p20187431.html
Sent from the HBase User mailing list archive at Nabble.com.
