Getting Off Of Profiler, A Twelve-Step Program

Wayne Sheffield has a blast from the past, repeating an old T-SQL Tuesday to show how to use Extended Events:

Now that you have this XE session scripted out, it can be easily installed on multiple servers. If you encounter a deadlock problem, you can easily start the XE session and let it run to trap your deadlocks. They will be persisted to a file dedicated for the deadlocks. You can use my Deadlock Shredder script at http://bit.ly/ShredDL to read the deadlocks from the file and shred the deadlock XML into a tabular output.

Note that the default system_health XE session also captures deadlocks. I like to have a dedicated session for just deadlocks. As lightweight as XE is, sometimes it may benefit a server to turn off the system_health session. Additionally, Jonathan Kehayias has a script that will take a running trace and completely script out an XE session for it. This script can be found at https://www.sqlskills.com/blogs/jonathan/converting-sql-trace-to-extended-events-in-sql-server-2012/. Even though this script is available, I like to figure things out for myself so that I can learn what is actually going on.

Extended Events are extremely useful for administrators, typically with a fraction of the overhead cost of  server-side (much less Profiler) traces.

Dark Queries

Michael Swart helps root out queries which get recompiled frequently and so won’t be in the cache:

Some of my favorite monitoring solutions rely on the cached queries:

but some queries will fall out of cache or don’t ever make it into cache. Those are the dark queries I’m interested in today. Today let’s look at query recompiles to shed light on some of those dark queries that maybe we’re not measuring.

By the way, if you’re using SQL Server 2016’s query store then this post isn’t for you because Query Store is awesome. Query Store doesn’t rely on the cache. It captures all activity and stores queries separately – Truth in advertising!

Click through for an Extended Event session which looks for recompilation.

SSAS Extended Events

Eugene Polonichko shows how to use Extended Events with SQL Server Analysis Services:

Cubes require frequent monitoring since their productivity decreases quite often (slowdowns during query building, processing time increment). To find out the reason of decrease, we need to monitor our system. For this, we use SQL Server Profiler. However, Microsoft is planning to exclude this SQL tracing tool in subsequent versions. The main disadvantage of the tool is resource intensity, and it should be run on a production server carefully, since it may cause a critical system productivity loss.

Thus, Extended Events is a general event-handling system for server systems. This system supports the correlation of data from SQL Server which allows getting SQL Server state events.

If you’re already familiar with relational engine XEs, this isn’t much of a stretch.

Replication Extended Events

Drew Furgiuele goes hunting for the most dangerous creature of all, replication-related extended events:

Extended events are great; they have all the goodness of profiler except you don’t use profiler. Win/win! More to the point, extended events let you quickly and easily view, sort, and aggregate events that occur on your instances. They also have powerful filters (really, a “where” clause) to limit noise. You have way more control over what you monitor, how you store the data, and how you view and use it. This makes them perfect use to track replicated transactions, since we want to measure at both an individual level and the aggregate.

I fired up management studio and went to “New Session” looking for some replication event goodness and I found…

… nothing. I tried looking for events that had even parts of the name replication in it. No such thing, apparently.

This doesn’t deter Drew and he ends up building some interesting events to infer the correct answers.

Retaining system_health Metrics

Taiob Ali looks at the system_health Extended Events session:

What this article does not tell you is your individual file size is 5 MB and number of maximum rollover file is 4. Meaning you will only get 20 MB of data. Once that limit is reached your older data is lost and there is no way to recover.

Often time you have an issue with the application pointing to SQL Server and you want to diagnose the problem with system_health files. You find that files already rolled over and information you are looking for is not available. I will explain how you can solve this problem and retain system_health session files for longer period.

Read on for the solution.

Invalid Database IDs In Extended Events

Jonathan Kehayias looks at a strange scenario:

I was recently emailed with a question about tracking database usage information using Extended Events and the person wanted to know why the event session was returning invalid database_id’s as a part of what it was collecting.  The funny thing is that I’ve seen this before during demo’s when I have a very specific concurrent workload running for what I am demonstrating and I used to make sure to do this exact thing so I could explain it during Immersion Events.  The TLDR answer to this is that Extended Events isn’t returning an invalid database_id at the time that the event occurs, the database_id just isn’t valid when the data is being reviewed or consumed.  How could that be you wonder?

The explanation turns out to be pretty easy, though not necessarily intuitive.

MAXDOP-Related Extended Events

Paul Randal looks at a couple of Extended Events which help you find CXPACKET waits and degrees of parallelism:

Now I’ll set up a simple Extended Events session to track down the offending code (based on the query from here). It’s very important that you query the sys.dm_xe_map_values DMV to find the correct number to use in the query for the CXPACKET wait, as these numbers often change from release to release, and even in Service Packs. For instance, CXPACKET was 191 in SQL Server 2014 RTM, but is 190 in the 2014 build I’m using.

Be very careful about running this in production, as the event will fire for *every* wait that occurs and so will likely affect your workload throughput, even though it’ll short-circuit if the wait isn’t CXPACKET. I’ll show you a better event to use lower down.

These are good events to know.

Azure SQL Database Extended Events

Arun Sirpal compares on-prem extended events to what’s available in Azure SQL Database:

There are 22 actions and 261 events. Naturally less than your local based SQL Servers, for example on my local 2014 machine running the above query returned 50 actions and 284 events.

There are a few subtle differences and a couple not-so-subtle differences, so it’s worth digging into if you plan to spin up an Azure SQL Database database.

UDFs And Recompilations

Erik Darling shows how to diagnose a high recompilation problem:

So yeah, that function seems to get up to something once for every ID you pass in. Remember that in our STUFF… query, we grabbed the TOP 10 each time. In the XE session, each time we call the proc, the string splitting function compiles and executes code 10 times. Bummerino. That’s the life of a loop.

On SQL Server 2016 (and really, with any non-looping code), we can get around the constant compilations with a simple rewrite. In this case, I’m calling 2016’s STRING_SPLIT function instead of the MSTVF function.

Read on for more.

Querying The Histogram XE Target

Kendra Little shows how to query the histogram target for Extended Events:

I like using the histogram target because it’s relatively lightweight — you can “bucket” results by what you’re interested in. In my case, I was interested seeing the cumulative number of file_read events by file name.

But there’s one problem: the histogram target is stored in memory, not in a data file. If you want to query that data and store it off in a table, it’s not obvious how to do that.

Click through to figure out how to do that.

Categories

April 2017
MTWTFSS
« Mar  
 12
3456789
10111213141516
17181920212223
24252627282930