A long time ago we talked about how to Import fixed width data with pgloader, following up on other stories still online at Postgres OnLine Journal and on David Fetter’s blog.
Back then, I showed that using pgloader made it easier to import the data,
but also showed quite poor performances characteristics due to using the
debug mode in the timings. Let’s update that article with
current pgloader wonders!
As presented at the PostgreSQL Conference Europe the new version of pgloader is now able to fully migrate a MySQL database, including discovering the schema, casting data types, transforming data and default values. Sakila is the traditional MySQL example database, in this article we’re going to fully migrate it over to PostgreSQL.
In our previous article about Loading Geolocation Data, we did load some data into PostgreSQL and saw the quite noticable impact of a user transformation. As it happens, the function that did the integer to IP representation was so naive as to scratch the micro optimisation itch of some Common Lisp hackers: thanks a lot guys, in particular stassats who came up with the solution we’re seeing now.
As I’ve been mentionning in the past already, I’m currently rewriting pgloader from scratch in Common Lisp. In terms of technical debt that’s akin to declaring bankrupcy, which is both sad news and good news as there’s suddenly new hope of doing it right this time.
About the only time when I will accept to work with MySQL is when you need help to migrate away from it because you decided to move to PostgreSQL instead. And it’s already been too much of a pain really, so after all this time I began consolidating what I know about that topic and am writing a software to help me here. Consider it the MySQL Migration Toolkit.
