My company requires a trip report after we go to a conference to talk about lessons learned and what we can bring back and use at the company. So I’m using this post to share with the community and to fulfill the company requirement.
I have my notes arranged day by day so I’ll take that approach for the post. At the bottom I’ll talk about the networking/community aspect of things I did at the summit.
Day 1 (Monday SQL in the City)
They were using continuous integration to identify issues as soon as possible. I used a similar type of system at a previous company and am working hard to get that in place at my current company.
They did mention a great idea I had not thought of before for unit tests. If you create a Db unit test and say it can’t take more than X seconds to run then you can have a simple way to check for performance problems. Couple key things, you’ll need an environment with consistent performance and the ability to control outside forces so you can validate this is a true performance change. Can be a great way for some simple performance tests though.
I enjoyed this presentation and he re-iterated many of the things I already knew about database design. I really liked how explained the different models and what should be in each model. My take away that I wrote down during the session was a role change we need to make to our current system.
“Users are Evil but they have checks” – Was my favorite line from this presentation.
Grant gave some great tips about performance tuning some of the key ones that I wrote down.
Have a baseline, this can be generated from the previous data over time so you can understand and define “Normal”.
Currently in my world I use qlikview to put my stats together and then compare hours to previous hours or days to previous days. Frequently I’ll compare this week to last week. For me this is how I typically define “normal” and try to compare what’s going on. I’ve recently added table size data into my reporting system so in the future I’ll be able to tell large growths and compare them to days without large growths.
Day 2 (Tuesday)
This was all PASS Meetings during the day and I’ll have a separate Blog post about that.
Day 3 (Wednesday)
Periodic Table of DMV’s
Tim did a great job of explaining the Dmv’s that were out in the system. The primary item I took away from this session was a need to monitor my Dmv’s closer particularly the Io Virtual File stats. I’ve had perfmon counters on my disk’s but in my current performance monitoring set I don’t get down to the detail provided in the IO virtual file stats and I need to add that. I’ve added a task on my list to get that done.
Profiler to Extended Events
Erin Stellato did an excellent presentation of migrating and working with some of the changes between Profiler and Extended events. I unfortunately had another PASS meeting and had to leave this session early but I did take some items away from it. I have lots of detail notes about how the extended events work so when I get to sit down and start working with them I’ll have all the info I need. We need to make a push to 2012 to really make this an easy change to extended events. We can and will work on this in 2008 but we will have to put a lot more time and effort into the change. We have a task on the books to migrate to 2012 in our organization. I need to check out the tools and GUI that Jonathan Kehayias has for 2008 so I can get started on our project now and then continue with it into 2012. We currently have a server side trace that always runs and collects the data we need so we have a very robust system to already do this but need to work on updating it now so we are ready for the changes.
Big Data for Business Intelligence Professionals
Jen Stirrup and Allan Mitchell did a great job of presenting the topic. This was a longer spotlight session as well it was 1.5 hours. I liked the original overall talk and discussion they started out with explaining big data. I think they missed the mark when they started to talk about PIG instead of hive first. PIG is a good tool to use but this audience is much more geared to sql queries and SQL ideas which hive fits into a lot better. They did circle around back to hive at the very end of the talk but I think many were confused by then. They showed HDinsight and what was available in the windows/hadoop ecosystem which is something I really wanted to understand since my Big Data implementation is using Linux and Cloudera so I wanted to understand the windows side. The console and commands looked basically the same just the OS was different. Key take aways that I have, Big Data is an approach to a problem. Think of it as Volume, Veracity, Variety, Veracity and Value. I have a feeling I’ll be writing a lot more about this in the future.
Why are things slow
Brent never fails to disappoint when it comes to a session. He showed us a new sp_blitz with some new features and what I really liked was the open source monitoring system in place at StackExchange. I really liked what I saw with the monitoring system and already started to get it installed here at the office. I intend to work on running the Blitz scipt more frequently as well to start adding it to my comparisons of what things looked like before and after releases of our code.
I spent a good amount of time in the community section and talking with many people about the community. Here is some of the notes from just general discussions.
- I explained frequently that I’ve gotten to a larger point with my community work that I formed a not for profit to handle everything and several people asked me about that process. TODO: blog post on this.
- Mentor programs were a big discussion and something I’ve been focused on for a long time. I’ve already picked up a domain for this and have worked on some documentation around it but I still have a lot of work to be done. Andy Warren shared a lot of great ideas on this as well and I think this could be a great and needed program. This is one of my highest focus’s currently that I’m working on in my spare time.
- SQL Saturdays were a huge discussion of course as well. I mainly listened in these discussion as I don’t have a lot of issues in running a SQL Saturday. I run events frequently and don’t have a ton of problems with my SQL Saturday. I want to see SQL Saturday expand past SQL and focus more on data and the ever changing data world but that’s a much larger blog discussion. TODO: Blog Post. J
- A discussion on my comment of “Done and Done” from this blog post/comment. This was not at the community tables but from a friend that I highly respect and intend to have a future blog post on. Not one that the friend will necessarily like but just because I respect the person doesn’t mean I’m going to agree with them. J