I’ve been doing a lot of work lately with kCura Relativity, a legal discovery tool with a SQL Server back end. I really enjoy working with George Orr, Scott Ellis, Mike Kolek, and the rest of the team at kCura because they’re fun people solving cool technology problems. It’s not too often that you hear a database guy saying, “Wow, I really love this third party software,” but I do. It’s not easy to build an ISV app that manages terabytes of data while letting complete strangers manage the database – and even add fields and indexes – so my hat is off to these guys.
If you’re a Relativity admin coming to Chicago for RelativityFest next week, say hi to me at the Advice Bar, where I’ll be hanging out in between classes. Bring your toughest database questions and we can play Stump the Microsoft Certified Master.
And yep, I’m attending classes too – I find the whole e-discovery business really interesting, and I like being able to speak the user language. I like being able to say, “I know where you’re clicking in the app that’s producing this database query, and here’s a more efficient way to get the data you need.”
3 Comments. Leave new
sweet…see you there!
the kcura guys definitely know their stuff (in fact, it was at RelativityFest last year that i first learned of brentozar.com). i’ve been working with Relativity for over four years now, and i agree with your assessment–i’ve worked with more than a few SQL Server-based applications in my time, and Relativity really stands apart.
i’ve had colleagues familiar with the application but with no prior SQL Server experience ask me for advice on how get started with SQL Server. my answer is always the same: “you’re in luck: you know Relativity, and Relativity is written the way a SQL Server application should be written.” after a quick tour through the schema, the basic concepts start to click, and fast. in addition to being an excellent piece of software, it has proved to be a valuable teaching tool.
Brent,
great talking to you at RelativityFest. I figured i’d ask here so that others might get the benefit of the answer. You mentioned that SQL’s default cost threshold for parallelism was too low, especially for Relativity Instances. Have you found a different value to be better for Relativity? Jonathan Kehiyias has some interesting info here:
http://sqlskills.com/blogs/jonathan/post/Tuning-e28098cost-threshold-for-parallelisme28099-from-the-Plan-Cache.aspx
but i’ll confess that after running this in our environment i’m not really sure how to use the data to come up with a “better” number.
qed – This is something that I’d rather work with my clients privately. If you’d like advice on it, feel free to email me at help@brentozar.com. Thanks!