The Art of War-Fast Track Data Warehouse & Fragmentation
When loading a Fast Track Data Warehouse it is important to ensure that your data is optimally laid out for Sequential I/O. Fragmentation is therefore the enemy.
Know your enemy.
Learn all about fragmentation - from the inside out. Let's get back to brass tacks and understand how SQL Server handles heaps and clustered indexes, re-visit page allocation and then apply this knowledge to our understanding of fragmentation. We will conclude by looking at some design patterns to help you minimise the impact of fragmentation on your next data warehousing project.
Key topics presented:
- Logical, extent and file fragmentation
- Heap and clustered index architecture and internals
- Page allocation - how it works and how to influence it
- Trace flags that help avoid fragmentation
- Fragmentation in action
- Design patterns for minimal fragmentation
James Rowland-Jones is a Principal Consultant for The Big Bang Data Company. His focus and passion is to architect and deliver highly scalable analytical platforms that are creative, simple, and elegant in their design. James specializes in big data warehouse solutions that leverage both SQL Server Parallel Data Warehouse and Hadoop ecosystems.
James is a keen advocate for the SQL Server community; both internationally and in the UK. He currently serves on the Board of Directors for PASS and sits on the organising committee for SQLBits - Europe's largest event for the Microsoft Data Platform. James has been awarded Microsoft’s MVP accreditation since 2008 for his services to the community.
You can find James on twitter @jrowlandjones