The issuing of a Current Time Increment, Cti, in StreamInsight is very definitely one of the most important concepts to learn if you want your Streams to be responsive. A full discussion of how to issue Ctis is beyond the scope of this article but a very good explanation in addition to Books Online can be found in these three articles by a member of the StreamInsight team at Microsoft, Ciprian Gerea. Time in StreamInsight Series http://blogs.msdn.com/b/streaminsight/archive/2010/07/23/time-in-streaminsight-i.aspx http://blogs.msdn.com/b/streaminsight/archive/2010/07/30/time-in-streaminsight... [More]

In this postI show how you can use StreamInsight to take events coming from SQL Server Profiler in real-time and do some analytics whilst the data is in flight.  Here is the solution for that post.  The download contains Project that reads events from a previously recorded trace file Project that starts a trace and captures events in real-time from a custom trace definition file (Included) It is a very simple solution and could be extended.  Whilst this example traces against SQL Server it would be trivial to change this so it profiles events in Analysis Services. &#... [More]

This tool is excellent when investigating what is going on in your StreamInsight Streams.  I was looking through the menu items recently and went to Query => Event Fields I found that there were a couple of columns not added by default to the event viewer (Reminds me of the fact that the Variables viewer in SSIS hides columns also) Latency NewEndTime EnqueueTime Here they all are together.     This gives us even more information about what is going on

Using the Trace Management objects API it is possible to capture the events happening on SQL Server or indeed Analysis Services.  I got to thinking that this would be a good thing to stream out in real time in StreamInsight and do some analysis “In Flight”.  So here is a short video of me doing just that.   Your browser cannot play this video. Learn how to fix this.

There are a number of ways to issues CTIs (Current Time Increments) into your StreamInsight streams but a quite useful way is to do it declaratively on your source factory like this public AdapterAdvanceTimeSettings DeclareAdvanceTimeProperties<TPayload>(InputConfig configInfo, EventShape eventShape) {     return new AdapterAdvanceTimeSettings(         new AdvanceTimeGenerationSettings(configInfo.CtiFrequency, TimeSpan.FromTicks(-1)),         AdvanceTimePolicy.Adjust); } T... [More]

Performing a LEFT OUTER JOIN in TSQL is easy.  The same logic in StreamInsight requires a little more work.  In this presentation I show exactly how you can perform a LEFT OUTER JOIN on your event streams. Your browser cannot play this video. Learn how to fix this.

Come join me for what promises to be a good evening.  18.8.2010.  Sign Up

Come join us at the UK SQL Server UG on 14.7.2010 in London

I will be tweeting on all subjects that interest me and come to mind.  Expect to see tweets about StreamInsight, SSIS, Data Mining and Spatial.  Follow me at @allanSQLIS

In this post I want to show a couple of ways to order the data that comes into the pipeline.  a number of people have asked me about this primarily because there are a number of ways to do it but also because some components in the pipeline take sorted inputs.  One of the methods I show is visually easy to understand and the other is less visual but potentially more performant.

I cannot help but think that this is currently wrong.  There are two editions of StreamInsight (SI), Premium and Standard.  A link to their differences is here Points to note Developer Edition and Evaluation Edition represent Enterprise Edition in SQL Server but not SI where they represent Datacentre Edition.  Won’t this be confusing when people evaluating SI find that it is not the same animal when they move to Enterprise Edition? If SI Premium is the only thing you want out of the SQL Server box then it is going to be a high cost. The latency rates as well quot... [More]

Her is an article I wrote for MSDN that helps explains the excellent viewer we get for Sequence Clustering models in SQL Server Data Mining.  I show you how the numbers you see are derived and also give an explanation as to what the icons you see in the viewer mean. Link to the article