This past weekend I attended SQL Saturday #149 in Minneapolis with my husband Adam, my brother Chris and another member of our local PASS chapter. Adam has his pilot’s license so we flew to Minneapolis the morning of the event. It was a beautiful and warm fall day – perfect weather for flying!
The local PASS chapter put on a great event! I was really impressed with the quality of the presentations. The event took place at the University of Minnesota; the venue was ideal for this type of an event.
The first presentation we attended was ‘Locks, Blocks and Deadlocks: Understanding How SQL Server Implements ACID Without Getting Burnt’ by Kyle Neier. Kyle did a nice job explaining various aspects of transactions and isolation levels via analogies to Black Friday shopping. He explained the differences between pessimistic (locking) and optimistic (versioning) concurrency control as well as explicit, implicit and automatic transactions. He also described the different types of isolation levels and scenarios when it is appropriate or inappropriate to use each type. He wrapped up the presentation by discussing how to detect and prevent blocking and deadlocks.
After lunch, we attended ‘Hadoop for the SQL Developer / DBA’ by Bill Preachuk. We all felt that we learnt a lot from this presentation and were happy we decided to attend. Bill provided an excellent explanation of the Hadoop distributed file system (HDFS) and map reduce algorithm. He described how data files are split into blocks and spread across the cluster. He also explained some of the fault tolerant features, including how the HDFS automatically replicates copies of the data to three different nodes and how the data is re-replicated to three nodes when a node fails. When queries are run in map / reduce jobs, they only operate on local data at the item level and the data is read and written in <key,value> pairs. Near the end of the presentation, Bill discussed the Hadoop Ecosystem which includes tools such as Hive, Pig, Sqoop, Flume and Mahout.
The next presentation was ‘Advanced Dimensional Modeling’ by Fazel Haris. The first part of his presentation included an introduction to dimensional modeling. After providing some basic background, Fazel covered more advanced topics of dimensional modeling such as Junk Dimensions, Fact Dimensions, Role Playing Dimensions, Referenced Dimensions and Many-to-Many relationships. Fazel walked through examples from an SSAS implementation in a healthcare environment.
Finally, we wrapped up the event by attending ‘Real-Time Analytics with SSAS Tabular DirectQuery’ by Paul Doyle and Nelson David. The presentation started with an overview of the Microsoft BI strategy and where various products align with respect to Individual BI, Team BI and Organizational BI. Paul then covered the benefits of DirectQuery, including up-to-data data with no processing required and support for large datasets. He also covered a few limitations including the lack of support for MDX queries, limitations of reporting clients (PowerView is the only option that is currently compatible with DirectQuery), lack of support for DAX functions, etc. The presentation include several good demos and examples.
I want to thank all the event coordinators, volunteers and presenters for making SQL Saturday 149 a success.