A trio of conferences
The trio of conferences started on the Sunday morning with the MVP Summit. This is Microsoft’s annual gathering of MVPs, where the company shows off new and upcoming features of its products and elicits comments, suggestions and criticisms from the MVPs. In the case of the SQL group, there’s often a lot of criticism! While everything discussed during the MVP summit is under NDA (which results in some amusing tweets during the conference), some of the highlights of the conference have since been announced publically and so can be discussed. These include:
- Sharding for Windows Azure SQL Database (http://blogs.msdn.com/b/frcogno/archive/2014/11/04/sneak-peek-horizontal-scale-using-azure-sql-database-elastic-scale-preview-day-1.aspx)
- Updatable, non-clustered columnstore indexes, including for Hekaton tables
- Stretching tables from a physical SQL installation to Azure SQL DB (http://sqlblog.com/blogs/rob_farley/archive/2014/11/06/pass-summit-keynote.aspx).
On the Monday, Red Gate ran the Seattle leg of its SQL in the City. These mini-conferences are a mixture of technical training and marketing material and there’s usually a good balance between the two. Tony Davis (editor for Red Gate Software) and I presented a technical session entitled ‘You did WHAT to my transaction log?!?!?’ The session aimed to show just how bad things can become if people try quick solutions to log growth, without bothering to understand how the log is used and how it works.
The annual PASS Summit started with a bang on the Wednesday morning, but fizzled out with the boring, marketing-focused keynote that followed. One of these days we will have technical keynotes on both days, but until then… As an aside, Chris Webb’s blog post on some of the reasons why the Wednesday keynote was badly received makes for an interesting and enlightening read.
As expected, nothing major was announced at the PASS Summit because we are halfway through a development cycle for SQL Server – no dates were given for vNext. There were some Windows Azure SQL Database announcements (unfortunately these did not include a shorter name for it), but they did include sharding and some hints as to the features that will be in vNext. These could incorporate updatable, nonclustered columnstore indexes and stretch tables to Windows Azure SQL Database.
Without much effort, the rest of Wednesday far exceeded the keynote in interest. Buck Woody discussed relevant data storage options, making the very, very important point that not everything should be shoehorned into a single solution just because that is the hammer. If part of an application’s data persistence fits naturally into a document database and half into a graph database, that is where they should go, and they shouldn’t both be crammed into something inappropriate.
Conor Cunningham, architect on the SQL team, introduced a feature for SQL Server vNext, called the Query Store. In a nutshell, Query Store is a persisted record of query performance, broken down by object and query and user-chosen interval. This feature will, in most cases, make it unnecessary to run trace/extended events sessions to capture a system’s query workload, because it will be done continually in the background. Its best feature is that since it’s persisted, a database backup will contain all the Query Store information with no additional work required.
Thursday started with a technical keynote from Dr Rimma Nehme, who works at the Microsoft Jim Gray Systems Lab. She gave an excellent technical presentation on cloud computing — for the record, she says we should blame the network people for the name ‘Cloud’. The rest of the day was just as good, with Paul Randal talking about advanced recovery options (including using hex editors on database files) and Bob Ward spending three hours talking about SQL Server IO.
By Friday, although many delegates were worn out, there was still a good-size audience for my session on the cardinality estimator. However, the later scheduling of the session could have been the reason for there being fewer questions and interaction than I would have liked. That said, all in all, the three conferences made another excellent week of technical training, friendship and learning.