email or call +1 888-618-9088

 

1209 Kinney Ave
Austin, TX 78704
USA

(512) 686-5300

We create content for clients to make complex subjects clear - videos, visuals, interactive and presentations.  

 

 

Blog

Posts with observations about visual communications, video, good writing, great design and more than a few technical notes. 

Graph Data & Tools Coolness

Matthew Dunn

We're working as part of a team - our client is helping a huge client of theirs with a very large-scale IT restructure.  Not naming names, but between the two of 'em there are well over 300,000 employees.  

The uber-client provided, among other things, a block diagram (PDF) of high-level IT resources - hardware, software and organization all mixed together.  250+ entities - some blocks-within-blocks.  Things like "Organization X Helpdesk".  BIG resources. 

Here's the punch-line. The block diagram was helpful, but it didn't really capture the relationships between those entities.  Adding connector lines wouldn't either. 

My brain did a bit of a short-circuit then, and said "Psst.  You.  Put down the coffee cup. Go dig up those articles on graph theory & graph databases from Evernote."   Which lead me (among other things) to this:

You may have heard of the 'social graph' in connection with Facebook.  You're friends with HIM, he's friends with HER, she LIKES this page. If you drew that on a whiteboard, you'd have balls-and-lines connecting all over the place.  That's one example of a graph.

That is, as it turns out, a very different kind of data problem.  What we call 'conventional' databases (relational database), and what we call "the world's greatest app" (Excel), pretty much stink at representing relationships like that.  Which means our client's client (that's kind of graph-data statement, isn't it?) is going to struggle to get an accurate understanding of all those IT resources. 

There are graph databases - this one, Neo4j seems to have a lot of momentum.  

I downloaded, played with it, read the (free) O'Reilly book on graph databases.  Seriously cool stuff. Seriously useful, too, I think.

Howevah...I suspect we won't end up getting to help the client's-client with their big understanding problem, using this powerful approach.  I'd love to. I think it would become a critical source of intel and even operational excellence. But...the culture there, which our contact point would have to battle, is sort of Powerpoint-and-Excel-solve-all-problems.  Sigh. Damn.   Excel rocks, but not for everything.  Powerpoint can be made to rock, but it can't fight its own nature, which is sequential, choppy, non-narrative and non-exploratory declarative fragmentation.  

Maybe somebody out there has a hairball graph-data kind of problem that they need help explaining. You call me, OK?

(Both of those graph links are open source - amazing what's out there!) 

 --md