Wouldn't it be great to be able to directly monitor specific behaviors within SQL Server, like, oh, I don't know, knowing exactly when, and how, someone is using BULK INSERT? Well, you can, thanks to Extended Events through the bulk_insert_usage event. Bulk_insert_usage The BULK INSERT command is extremely useful within SQL Server. It's a way to move data into the database and provide some formatting on the way, efficiently, all through T-SQL. Hard to argue with the utility. Obviously, if you're doing traditional data collection through Trace or Extended Events, you'll see BULK INSERT commands within the T-SQL. However, Extended Events provides a specific event that tracks just the behavior of BULK INSERT: bulk_insert_usage. Documentation on this is somewhat sparse. Some of the best is from a standard source, Jason…
I hear this one all the time: How do I find out who implemented object changes? I also get: Can I see the query that caused object changes? Let's take a look at how you might audit who is doing what and how to your databases. Object Changes in Extended Events If you open up the New Session window for Extended Events in SSMS, the easy way to track down events is to simply type into the box. Here, we care about capturing object changes, so I'm going to simply type object, then scroll a bit: There we are object_altered, object_created and object_deleted. These are the same events that you would see in Trace. Let's use the GUI and take a quick look at what fields they capture: That's in…
One of my favorite little bits of information about Extended Events is the fact that everyone running a full instance of SQL Server has deadlock information available to them, even if they never enabled Trace Flag 1222 through the system_health session. That captures the xml_deadlock_report which has the full deadlock graph. However, what if you want to capture deadlock info, but, you're dealing the GDPR, and transmitting query values could be problematic? Enter xml_deadlock_report_filtered. xml_deadlock_report_filtered If you do a search for this event, you're not going to find much. Doesn't seem like anyone, including Microsoft, has bothered to document it. This is not going to be a comprehensive definition for all things xml_deadlock_report_filtered. However, I can show you why you might want to use it. This is a port of…
Most of the time when I talk about or demo Extended Events, I spend more time talking about query tuning (I have a problem). However, there are tons of things that you can do with Extended Events. Here's a little one that came up, auditing unique constraint violations. Unique Constraint Violations Whether we're talking a primary key or just a constraint, the error you get is number 2627 when you attempt to add a non-unique value. So, the code for a simple way to track this in Extended Events would look like this: CREATE EVENT SESSION [UniqueConstraintViolation] ON SERVER ADD EVENT sqlserver.error_reported (WHERE ([error_number] = (2627))); That's it. That's all you need. Probably, it'd be a good idea to output this to a file (that's usually what I do). However,…
Probably the single most important factor when deciding which query to tune, or actively tuning a query, is how you go about measuring query execution time. SQL Server provides a number of different mechanisms (really, maybe too many) to get this done. However, all measures are not created equally. In fact, they frequently disagree with one another. Let's take a look at this odd phenomenon. Measuring Query Execution Time Before we get into all the choices and compare them, let's baseline on methodology and a query to use. Not sure why, but many people give me blow back when I say "on average, this query runs in X amount of time." The feedback goes "You can't say that. What if it was just blocking or resources or..." I get it.…
In Azure SQL Database for quite some time and now available in SQL Server 2017, Microsoft has put a lot of the knowledge they've gleaned from running more databases that any of the rest of us ever will to work with Automatic Tuning. Automatic Tuning The core of automatic tuning at this point in time (because I'm sure it's going to evolve) is the ability of the query engine to spot when a query has generated a new plan and that new plan is causing performance to degrade. This is known as a regression in the plan. It comes from bad parameter sniffing, changes in statistics, cumulative updates, or the big notorious one, the cardinality estimator introduced in SQL Server 2014 (it's been almost four years, I'm not calling it…
This requires an immediate caveat. You should absolutely be using sp_executesql over any type of non-parameterized execution of T-SQL. You must parameterize your T-SQL because the lack of parameters in building up and executing strings is a classic SQL Injection attack vector. Using straight ad hoc T-SQL is an extremely poor coding choice because of SQL Injection, not because there is something that makes one method faster than the other. Yet, I see in performance checklists that you should be using sp_executesql over straight ad hoc T-SQL because it will perform faster. That statement is incorrect. Some Discussion Let me reiterate the caveat before we continue. I 100% advocate for the use of sp_executesql. This function is preferred over ad hoc SQL because, used properly (and isn't that usually one of the main problems,…
It really is so much easier to just throw hardware at badly performing databases. Just buy a bigger, faster server with more and faster disks and you can put off doing tuning work for another 6-9 months, easily. But, for most of us, sooner or later, our performance problems get so big or, we just don't have any more money to spend, and we're stuck. We have to tune the queries. And frankly, query tuning is a pain in the nether regions. But, after you've tuned queries 20 or 30 times, you start to recognize the patterns and it gets easier (never easy, just not as hard). But, if you haven't done it 20 or 30 times, what do you do? My suggestion, talk to someone who has done it…
Tom LaRock has a new meme for Meme Monday. It’s all about the problems caused in your system other than disks. Thankfully, despite the title, I don’t have to list 99 separate things, only 9, but you know what, 99 is possible. I’m going to present the problems. You find the solutions on your own today. Let’s go. Recompiles I’ve seen queries so big that they take more than three minutes to compile. That’s the edge case, but as an edge case it is educational. The most important thing to remember about recompiles is that they are driven by data changes. Once a threshold is reached on any given set of statistics, all queries referencing that set of statistics gets marked for recompile. The key words and tricky phrase here…
I have to share this one. I'm working on a set of queries, tuning them. They're all following a similar pattern. They're passing in XML data sets which are then shredded into a temporary table. Once that's done, the temp table is used in a series of joins to other tables to return the data. Some of the queries were missing indexes, some were structured a bit poorly, but overall, it was pretty standard stuff. Until... I ran a query that looked, at first glance, the same as the rest. When I looked at the execution plan, I saw a warning symbol, one of those little exclamation points on an operator. I figured, based on the other issues with this database, that it was just some out of date or…