Graph databases are great. But if I read more FUD about RDBMSs, I’m going to have another flashback to the 80’s where someone is telling me about how you don’t want to dismantle your car every time you park it in the garage (OODBMS reference for the oldies).
I recently read a graph-database book that explained that every SQL join requires computing a cartesian product. And yesterday, a post that explained that equijoins have an exponential order of complexity. (I guess… if the exponent is 1.) It’s getting very frustrating to read through the inaccuracies and FUD, and now, to me, it is all just turning into
Which is why it was nice to see @guyharrison‘s even-handed and accurate treatment in his new book Next Generation Databases, and then to see this on LinkedIn (thanks to Shahram Ghandeharizadeh at USC for posting it and @markcallaghan for tweeting it). I’m reading the paper that used Neo4j and SQL-X first. I look forward to learning more about the many forms of data storage, access and usage, based on real research and data. Graph DB writers: don’t ruin your case by turning it all into “joins are bad, mkay” kinds of regurgitations.
