Recently a coworker asked me this question:
Should I expect variance between minutes and hours for the same query?
And I was forced to give him this answer:
Potentially, but not commonly. Query planning is an inexact science, and regardless of the query being the “same query,” the data is not the “same data.” This isn’t generally the case, but on occasion, changes in data can affect the query execution path.
It’s about the time for year-end performance reviews. While I’m always afraid I’ll narrowly avoid being fired for gross incompetence, that’s not usually how it goes. But that meeting did remind me about a bit of restructuring I plan to impose for 2017 that should vastly improve database availability across our organization. Many of the techniques to accomplish that—while Postgres tools in our case—are not Postgres-specific concepts.
Much of database fabric design comes down to compromise.
One of the best features Postgres boasts is the ability to adapt. Any schmo off the street can write an extension and bolt it onto Postgres with nary a second glance. As proof, I’m going to whip one up really quick. That should be enough to convince anyone that it takes no skill at all to add functionality to Postgres.
Just so our extension actually does something, let’s start off with the instant-runoff code we wrote a few weeks ago.
The Postgres query planner is house of cards built upon the ever-shifting sand of our data. It has the utterly impossible mission of converting our ridiculous and inane requests into a logical series of fetch, filter, sort, join, and other instructions. Then the resulting steps must be ruthlessly efficient or the execution phase could very well saturate every hardware resource available; Set Theory isn’t very forgiving.
Forewarned is forearmed is very apt when applied to database query planners.
The United States held an election recently, and there has been some … mild controversy regarding the results. Many raised issues about this before the election itself, but what if we had used instant-runoff voting instead? More importantly, can we implement it with Postgres?
Well, the answer to the last question is a strong affirmative. So long as we don’t break the results down into voting districts, and make wild unsupported assumptions regarding rankings, that is.
