Re: RDF Storage Systems

On 2001-12-03 4:09 AM, "David McCusker" <david@treedragon.com> wrote:

> I wrote a lengthy reply you can link at my weblog here:

Thanks so much. For you and Mathew, here are my two key questions:

1) For our initial versions, what free database should we use?
[ I'm imagining a three-month time frame for a prototype. ]

2) In a dbm which only exports a dict, how do we store RDF triples?
[ Ideally, we'd not have to duplicate a lot of data. ]

These are the problems we need to solve, I think. I'm interested in your
recommendations.

A full response to David is below.

> http://www.treedragon.com/ged/map/ti/newDec01.htm#02dec01-scaling
>
> ? I'll start with answering a semi-public email Aaron sent me.
> ? (I'm much happier being able to write for a public mode.)

Same here. I like sharing with others, as you do.
Plus, building an archive is helpful later on.

> Aaron is considering size figures more on the order of 60GB.
> This is rather big as content goes these days. A big disk size.

Indeed. I was a rather surprised when Mathew gave me that number.
He did say he'd be willing to take it in 2GB chunks.

> A 60GB database sounds a bit like all eggs in a basket to me.

I left out some context. The DB is part of a larger system.
It's part of a network, where other nodes are storing that data too.
 
> The idea is to jump from 100K files to 1GB near painlessly.

This sounds very useful for our prototypes. Or simple clients.
 
(BTW, some pages on your site are out of date. Your index is awfully old:

http://www.treedragon.com/ged/map/ti/time.htm

The header here still says Oct01:

http://www.treedragon.com/ged/map/ti/newDec01.htm

Writing in short sentences like this is fun.)

-- 
      "Aaron Swartz"      |               Swhack Weblog
 <mailto:me@aaronsw.com>  |   <http://blogspace.com/swhack/weblog/>
<http://www.aaronsw.com/> |      something different every day

Received on Monday, 3 December 2001 17:34:14 UTC