Press "Enter" to skip to content

Category: Administration

Workload Classification with Resource Governor in Azure Synapse Analytics

Niko Neugebauer keys in on an interesting addition to Azure Synapse Analytics:

Given that we can specify 5 different parameters (USER MEMBERNAME, ROLE MEMBERNAME, WLM_LABEL, WLM_CONTEXT, START_TIME/END_TIME) – there must be a prioritisation mechanism in order to decide which condition gets selected. This mechanism is called Parameter Weighting in Azure Synapse and it assigns the following weight to each of those parameters:
USER = 64
ROLE = 32
WLM_LABEL = 16
WLM_CONTEXT = 8
START_TIME/END_TIME = 4
meaning that if the Workload Classifier fits into the timeframe START_TIME/END_TIME, WLM_LABEL & ROLE – it will receive 52 points = 4 + 16 + 32,
while a different Workload Classifier that fits into WLM_CONTEXT & USER will get 72 points = 8 + 64, thus will prevail and will be selected over the first Workload Classifier.

Azure Synapse Analytics (including when it was known as SQL Data Warehouse) has had some resource governor-related things I’ve wanted in the box product for a while, including labels (which are better than using application name).

Comments closed

Queries Using tempdb: a Whodunit

Dave Bland shares a database detective story with us:

To find the culprit, we tracked down an old informant named sp_who2.  At first he had no comment, but we knew he had some information that could help us, so we kept after him.  Finally, he grew tired of us and gave us something just to make us go away. He said we needed to talk to his associate sysproccesses.  At first this guy was hard to find, but we located him sitting at a table.  He wasn’t really doing anything, just sitting there. So we approached him and sat down to see if we can get what we are looking for.  He also had no comment and referred us to his lawyer, sys.dm_exec_sessions.  After some searching, we were able to catch up with him at the DMV and he was not in a good mood.  He gave up some information, just not exactly what we are looking for.

Dave shows how we can figure out who created a specific temp table (a global temp table, in this case), the query that account used to create the temp table, and the time the temp table was created.

Comments closed

Deleting Packages from the SSIS Catalog

Mala Mahadevan performs important cleanup work:

I will be blogging on a few things I learn on my journey to SSIS expertise. This is the first one. This came about as a result of wanting to delete a few packages from ssis catalog. We do not use these packages any more, and I wanted to clean them out from the project which resides on a few servers. I looked into a few ways of doing it.

Click through for three methods, including an in-depth discussion of the third (and least obstructive).

Comments closed

A Naming System for Schedules

Daniel Janik shares a naming scheme for schedules in systems like SQL Agent and Azure Data Factory:

This tip comes from my DBA days working with SQL Agent Job schedules. If you’ve ever worked on a server where many people created job schedules you’ll know exactly what I mean when I say the schedule names can be really annoying.

This is because the names are not meaningful at all. They are either a GUID thanks to SSRS or something useless like “Schedule 1” or you have 6 different versions of “Every 5 min” when the schedule actually only runs every 15 min on Mondays.

The Linux nerd in me says “Could’ve just used cron naming.” I think Daniel’s naming scheme takes a little bit of time to get used to, but it makes sense.

Comments closed

The Benefits of Indirect Checkpoints

Aaron Bertrand has success with using indirect checkpoints:

I was a bit perplexed by this issue, since the system was certainly no slouch — plenty of cores, 3TB of memory, and XtremIO storage. And none of these FlushCache messages were ever paired with the 15 second I/O warnings in the error log. Still, if you stack a bunch of high-transaction databases on there, checkpoint processing can get pretty sluggish. Not so much because of the direct I/O, but more reconciliation that has to be done with a massive number of dirty pages (not just from committed transactions) scattered across such a large amount of memory, and potentially waiting on the lazywriter (since there is only one for the whole instance).

Read on for several links and the results of Aaron’s testing.

Comments closed

Using Powershell to Configure Database Mail and SQL Agent Alerts

Eric Cobb shows us how to use Powershell to set up database mail and SQL Agent alerts:

As a DBA, you need to know when there’s a problem on your SQL Servers. And while I highly recommend you use a full-fledged monitoring system, there are also some things you can set up on your SQL Servers so that they will tell you when certain things go wrong. This doesn’t replace a full monitoring system, but setting up the below alerts will give you notification when SQL Server encounters things like corruption or resource issues.

Even with a full-fledged monitoring system, there are places where you can still make use of mail and side alerts.

Comments closed

Stored Parameter Procedure Caching

Greg Dodd follows Betteridge’s Law of Headlines:

When SQL Server caches your plan, it caches it with the parameter values that you pass through the first time, and it assumes that the same query plan will be the best one for any parameter you pass in next time.

But does SQL Server always cache your parameters? Does it always keep track of what you pass in?

Click through for a demonstration good enough to give you a conclusive answer.

Comments closed

The Problems with Shrinking tempdb Data Files

Andy Mallon explains why you should never shrink tempdb data files:

I recently wrote about growing, shrinking, and removing tempdb files. In that article I explained that SQL Server won’t move a page that contains an internal worktable object, and thus trying to shrink tempdb files can be futile. Today, I’m going to explain how attempting to shrink tempdb files can actually be harmful.

Andy has good advice for tempdb here. Shrinking other database is generally bad but sometimes necessary; shrinking tempdb can lead to all kinds of problems.

Comments closed

Accessing Managed Instances from SSMS

James Serra shows us what we need to do in order to reach an Azure SQL Managed Instance from SQL Server Management Studio:

It used to be that the only way to use SQL Server Management Studio (SSMS) against Azure SQL Database Managed Instance (SQLMI) was to create a VM on the same VNET as SQLMI and use SSMS on that VM. That VM was usually called a jumpbox (see instructions here).

But about a year ago Microsoft added a way to use SSMS without using a VNET (announcement) by allowing you to enable a public endpoint for your SQLMI. This made it easy for me to access a SQLMI database on my laptop.

That change enables what James shows us.

Comments closed