Abstract: Microsoft’s Synapse DB is based on a Massively Parallel Processing (MPP) architecture. When your DB is designed right, all that compute and parallel workload can get answers to your queries fast. Designed wrong and a lot of that MPP power is used compensating for our design mistakes. This session will have an overview of the MPP architecture and how it influences design decisions around table types and storage type. Then we will review some best coding practices for Synapse DB and see how for large Upserts a CTAS (create table as select) and partition switch is the way to go. Lastly, since despite our best efforts we will write some poor performing code, we will take a look at troubleshooting and tuning techniques and how they were used in some real world examples.
Speaker Bio: Ted Tasker is a Data Architect at Insight Digital Innovations. Ted started working with SQL Server 4.21 in 1993 and has had a career centered on SQL Server ever since. The release of SQL 7 / OLAP shifted Ted’s focus to data warehousing and he led a team in deploying Disney Online’s first DW and OLAP reporting solution. After years of leading DW and BI teams, Ted joined Microsoft and was the technical lead for the first sale in the world of Microsoft’s PDW (Parallel Data Warehouse). He returned to consulting with a focus on Microsoft’s MPP database platforms and continued to work with PDW as it became APS, then Azure SQL DW and now the relational db platform for Azure Synapse.
Monday, April 13, 2020 – 6:00 PM to 7:00 PM EDT