Accumulo + OODBMS = ?

[Benson Margulies <bi...@gmail.com>]

Once upon a time, a long, long time ago, I was one of the implementors
of ObjectStore, now a product of Progress Software. I found myself
musing this morning about the possible intersection of giant scale and
OODBMS. Have any of you 'accumul-ators' considered some sort of object
mapping scheme? Or is there one that I've been too obtuse to notice?
Or is it just a dumb idea?

Re: Accumulo + OODBMS = ?

[Aaron Cordova <aa...@cordovas.org>]

Someone on the Accumulo team in the past may have overlaid JPA onto Accumulo ... or maybe Java Objects ... one of those. But I don't think the query language was there ..

I'll post the code I was using as soon as I track it down - it was also a demonstration of using another JVM language over Accumulo, in this case, Jython.

On Jan 24, 2012, at 10:01 AM, Benson Margulies wrote:

> On Tue, Jan 24, 2012 at 9:07 AM, Aaron Cordova <aa...@cordovas.org> wrote:
>> Such a thing might be possible, but I don't think anyone on the accumulo team is very familiar with OODBMSs - can you recommend some resources to get caught up?
> 
> You have homed right in on the big questions of what an OODBMS is, anyhow.
> 
> ObjectStore and it's competitors strive, as you write, to more or less
> transparently persist full Java objects, mapping collection types to
> index data structures and then adding a query language to traverse
> them. Mapping this sort of thing onto Accumulo would be a gigantic
> effort. It might be interesting to see of OrientDB could be married to
> Accumulo.
> 
> The weaker cousin of all this is the ORM notion, which is just what
> what you describe below. I'm in the process of trying to decide where
> to go next with a pile of code using an RDBMS plus an RDF triple
> store, and Accumulo is one alternative that I want to consider. Is
> that code somewhere that a person could goof around with?
> 

Re: Accumulo + OODBMS = ?

[Benson Margulies <bi...@gmail.com>]

On Tue, Jan 24, 2012 at 9:07 AM, Aaron Cordova <aa...@cordovas.org> wrote:
> Such a thing might be possible, but I don't think anyone on the accumulo team is very familiar with OODBMSs - can you recommend some resources to get caught up?

You have homed right in on the big questions of what an OODBMS is, anyhow.

ObjectStore and it's competitors strive, as you write, to more or less
transparently persist full Java objects, mapping collection types to
index data structures and then adding a query language to traverse
them. Mapping this sort of thing onto Accumulo would be a gigantic
effort. It might be interesting to see of OrientDB could be married to
Accumulo.

The weaker cousin of all this is the ORM notion, which is just what
what you describe below. I'm in the process of trying to decide where
to go next with a pile of code using an RDBMS plus an RDF triple
store, and Accumulo is one alternative that I want to consider. Is
that code somewhere that a person could goof around with?



>
> I have, in the past, prototyped an implementation of a object mapping for Accumulo similar to that provided in Google's AppEngine. It allows users to define subclasses of a Model class whose fields can be serialized to the database and retrieved via a simple query language.
>
> But if the intent of an OODBMS is to use the database to follow object references, I don't think anyone has attempted grafting such functionality over Accumulo yet, but it would be an interesting exercise to try to do so.
>
> On Jan 24, 2012, at 8:55 AM, Benson Margulies wrote:
>
>> Once upon a time, a long, long time ago, I was one of the implementors
>> of ObjectStore, now a product of Progress Software. I found myself
>> musing this morning about the possible intersection of giant scale and
>> OODBMS. Have any of you 'accumul-ators' considered some sort of object
>> mapping scheme? Or is there one that I've been too obtuse to notice?
>> Or is it just a dumb idea?
>

Re: Accumulo + OODBMS = ?

[Aaron Cordova <aa...@cordovas.org>]

Such a thing might be possible, but I don't think anyone on the accumulo team is very familiar with OODBMSs - can you recommend some resources to get caught up?

I have, in the past, prototyped an implementation of a object mapping for Accumulo similar to that provided in Google's AppEngine. It allows users to define subclasses of a Model class whose fields can be serialized to the database and retrieved via a simple query language. 

But if the intent of an OODBMS is to use the database to follow object references, I don't think anyone has attempted grafting such functionality over Accumulo yet, but it would be an interesting exercise to try to do so.

On Jan 24, 2012, at 8:55 AM, Benson Margulies wrote:

> Once upon a time, a long, long time ago, I was one of the implementors
> of ObjectStore, now a product of Progress Software. I found myself
> musing this morning about the possible intersection of giant scale and
> OODBMS. Have any of you 'accumul-ators' considered some sort of object
> mapping scheme? Or is there one that I've been too obtuse to notice?
> Or is it just a dumb idea?