Database subsetting. What it is, what it’s for, and why we’d love to hear from you.

We’re looking into database subsetting, and if you need it or are investigating it, we want to hear from you. Please share your thoughts in the comment section below.

What is database subsetting?

Database subsetting is creating a copy of a database that contains only a portion of the data, while still being referentially intact.

For example, suppose you have a table in a database containing millions of customers, but you only want to work with a database that contains thousands. Simple enough, just copy over a fraction of the rows, right?

But then suppose you also have an orders table that contains references to the customers. You’ll need to take only those orders that are related to customers in your sample. As your schema grows more complicated, so does the process of taking a valid subset of the data, and the need for automating the process becomes greater.

Why would I want a subset of my data?

We think there are many reasons organizations need subsets of data for development work, and right now we’re looking at three use-cases:

Testing, development and training environments – developers often want to work with real data, but providing a copy of your entire production database would be too costly in time and disk space, especially if you need to provide a copy for each developer. Subsetting allows you to work with data that contains all the necessary links between tables for your programs to function, but for a fraction of the cost.

Multiple organizations or departments – you may have a database that contains data for a number of different organizations or departments, and providing the whole database would expose data to individuals without the correct permissions. Instead, you can provide a subset containing just the data that is relevant to a particular organization or department.

Old data – you may want to discard a portion of your data based on some criteria. For example, you might want to clear out data older than a certain date, or data which you can no longer keep in order to comply with regulations like the GDPR.

What do you think?

Do you agree with us? Can you think of any other scenarios where you might need database subsetting? Are there other aspects of the use-cases we’ve outlined that should be considered?

We want to gain a deeper and broader understanding of the demand for data subsetting, so if your business has a need for any of the above or you would be interested in data subsetting for any other reason, please share your thoughts in the comment section below.

Share this post.

Share on FacebookShare on Google+Share on LinkedInTweet about this on Twitter

Related posts

Also in DevOps

The march of DevOps from the backroom to the boardroom

By encouraging collaboration and teamwork, removing the barriers between development and operations, and introducing automation, DevOps speeds up software delivery and enables features to move from th...

Also in Database development

Why you should automate your release pipeline

Release Day used to mean weekends spent deploying the latest code, database changes, and website content. Release teams would huddle together in war rooms troubleshooting issues, deploying hotfixes, c...

Also in Blog

Why code quality is vital in the world of database DevOps

Everyone understands the importance of code quality for applications, particularly when DevOps results in releases becoming faster and faster, reducing the room for error. The same issues increasi...

  • Pingback: Dew Drop - June 21, 2018 (#2750) - Morning Dew()

  • Markoolio

    One reason to work with a subset of data is data protection requirements. You may use outsourced developers who don’t have high enough security clearance to access all the data.

  • IanB

    We use it at my company for testing purposes. There is no replacement for having representative data in your test/dev environments as it will highlight data patterns that you did not anticipate. Also useful to put real data through the battery of automated tests to make sure they still pass with real world data.
    Our biggest problem was not chasing down foreign key restrictions but generating the subset in a reasonable amount of time as we have to reduce about 22TB of data down to about 4TB.

  • Bri-Guy

    Didn’t Redgate buy just such a tool (www.databee.com) a few years ago? I am interested in data subsetting for development and QA databases because running dev/QA activities on full terabyte-sized copies of production databases can take a very long time.