Press "Enter" to skip to content

Author: Kevin Feasel

Query Store Checks in dbatools

Published 2020-11-25 by Kevin Feasel

Jess Pomfret contributes to dbatools:

Once I was happy with my settings, I realised we were missing a ‘test’ command for dbatools. The suite of ‘test’ functions in dbatools (a lot that end up as checks in dbachecks btw!), give us an easy way to check our environment against best practices, or our desired settings.

Since dbatools is open-source I was able to write this function (Test-DbaDbQueryStore) and get it added into the module. It’s included as of version 1.0.131, so make sure you’re up to date.  Taking Erin’s suggestions and wrapping them in a little PowerShell, I can make it easier for myself and everyone else to make sure we’re following her guidelines.

Click through to see what those settings look like and how you can compare against current settings.

Comments closed

Creating Sequence Diagrams for SSIS Packages

Published 2020-11-25 by Kevin Feasel

Aveek Das has an idea for documentation:

In this article, I am going to explain in detail how to document SSIS packages using Sequence Diagrams and the importance of these diagrams in the field of software engineering, no matter which programming language are you using. In my previous article, I have talked about the various UML Diagrams that are being used to document various software engineering processes. Also, I have talked about modular ETL architecture and how to create such a modular package in SSIS. Sequence diagrams are also a part of the broader UML Diagrams which define the interaction between the various components in the system in a chronological manner.

My gut feeling is that this works best with medium-sized collections of packages, where we’re talking 10-30 or so packages in total, and that for something much larger, I’d want an automated tool to build diagrams for me. But I could be way off base on that.

Comments closed

Contingent Power BI Dataset Refreshes

Published 2020-11-25 by Kevin Feasel

Chris Webb has an interesting problem to solve:

This week a customer came to me with the following problem: they had scheduled the refresh of their dataset but their source data wasn’t always ready in time, so the old data was being loaded by mistake. The best solution here is to use some kind of external service (for example Power Automate) to poll the data source regularly to see if it’s ready, and then to refresh the dataset via the Power BI REST API when it is. However, it got me thinking about a different way of tackling this: is it possible to write some M code that will do the same thing? It turns out that it is, but it’s quite complicated – so I don’t recommend you use the code below in the real world. Nevertheless I wanted to write up the solution I came up with because it’s interesting and you never know, it might be useful one day.

Read on for the less-than-optimal solution, but do check out the better solution Chris describes.

Comments closed

Global Parameters in SSIS Framework

Published 2020-11-25 by Kevin Feasel

Andy Leonard has an update for us:

I’m happy to announce the latest version of our SSIS Framework includes global parameters! I can hear some of you thinking, …

“What Are Global Parameters, Andy?”

I’m so glad you asked! SSIS ships with package-scoped and project-scoped parameters. Project-scoped parameters may be used in any SSIS package in the project; package-scoped parameters are only available within the context of a single SSIS package. This functionality reduces repetition in SSIS package development and execution configuration.

Global parameters allow our SSIS Framework customers to set parameters and values that apply to the entire SSIS Catalog.

Now that you know what they are, Andy has an example of them in action. Global parameters aren’t part of the community edition, but they do look interesting.

Comments closed

Defining Simple Regression

Published 2020-11-24 by Kevin Feasel

Vincent Granville has a new algorithm for us:

The model-free, data-driven technique discussed here is so basic that it can easily be implemented in Excel, and we actually provide an Excel implementation. It is surprising that this technique does not pre-date standard linear regression, and is rarely if ever used by statisticians and data scientists. It is related to kriging and nearest neighbor interpolation, and apparently first mentioned in 1965 by Harvard scientists working on GIS (geographic information systems). It was referred back then as Shepard’s method or inverse distance weighting, and used for multivariate interpolation on non-regular grids (see here and here). We call this technique simple regression.

Read on to learn more about simple regression, including how to implement it and how it performs.

Comments closed

Automating Database Deployments: Why Not?

Published 2020-11-24 by Kevin Feasel

Grant Fritchey asks a question:

Building out processes and mechanisms for automated code deployments and testing can be quite a lot of work and isn’t easy. Now, try the same thing with data, and the challenges just shot through the roof. Anything from the simple fact that you must maintain the persistence of the data to data size to up time, and you have real problems in front of you.

However, adopting database deployment automation and testing has enormous benefits. Faster, safer, production deployment enhances the protection built around your production systems. Whether we want to use the loaded term of DevOps or not, the benefits of this style of development and deployment are easily documented and measured.

So, why are so few people doing it?

Grant gives some of the outline and lays out one response. I am seeing a lot more automation over time, but one underappreciated facet in this is a lack of trust for automated processes from humans. I think a good percentage of DBAs don’t trust that the automated process will get things correct, especially when dealing with complex chains of dependencies. An automated process may be less likely to make a mistake in a step, but it will also be unable to reason through an ambiguity and could perform an undesirable action in the event of unexpected circumstances. That’s a pretty big risk for DBAs who are concerned about their data. I can see a few other reasons as well, but this is one which I don’t hear often enough in these discussions.

Also, Grant asks people to fill out the State of Database DevOps survey, especially those people who are not automating database deployments.

1 Comment

Creating Custom T-SQL Snippets for SSMS

Published 2020-11-24 by Kevin Feasel

Dave Mason follows up on yesterday’s post:

I have a number of scripts and queries I’ve written and curated over the years. They’re not organized that well, scattered across different folders. Some are named poorly or grouped in a questionable manner. There are a handful that I tend to use the most. And yet with that small number, I sometimes have difficulty quickly finding a particular script (if I can find it at all), or spending too many mouse clicks to find it. It dawned on me recently to make use of code snippets.

Code snippets may have been intended primarily to aid code writing, but they can assist with administrative tasks too. 

These can provide a considerable benefit for data platform specialists.

Comments closed

Making Life Harder for Filter Operators

Published 2020-11-24 by Kevin Feasel

Erik Darling has a how-not-to guide for us:

We looked at a couple examples of when SQL Server might need to filter out rows later in the plan than we’d like, and why that can cause performance issues.

Now it’s time to look at a few more examples, because a lot of people find them surprising.

Read on for several examples. Also, because the bribes came through I don’t mind shilling for Erik, check out 25 hours of recorded content for $100. I think Erik’s knowledge is worth at least $4 an hour. Maybe even $5.

Comments closed

RANKX on Multiple Columns in DAX

Published 2020-11-24 by Kevin Feasel

Alberto Ferrari walks us through ranking based on multiple columns:

DAX offers the RANKX function to compute ranking over a table, based on measures or columns. One limitation of RANKX is that it is only capable of ranking using a single expression. Oftentimes it is necessary to use multiple columns to obtain a ranking, either because the business requirement dictates it, or because you want to rank ties with different criteria.

As a demonstration, we rank customers based on their purchase volume. To artificially introduce ties, we use the Rounded Sales measure, that rounds the sales amount to the nearest multiple of one thousand. Using Rounded Sales, several customers show the same amount of 10,000.00. Because they are ties, their ranking must now be defined by alphabetical order based on their names.

Read on for two methods to solve this problem.

Comments closed