Designing for Performance Part 4

[Chris Pearson]

Continuing where we left off with our designing for performance series, let's discuss a few common pitfalls that lead to poor performance. These come straight from a SN webinar delivered by their Account Escalation Engineering team.

The first of these pitfalls is the lack of proper indexing on tables you are querying. Any field which you are using in your query which isn't indexed can have extreme consequences on the performance of your code. SN now helps us out when creating transform maps which is very nice, prompting you to index any columns you've selected to coalesce on. If you're running import set API integrations in your instance, it might make sense to ensure you've indexed those columns properly.

The team used an example of a real customer dealing with multiple P1 incidents and service interruptions. They had implemented a custom table which was small...only 180,000 rows of data on the table. They were querying the table repeatedly, but didn't have any indexing on the table. The query time for these transactions where taking 9 seconds on average. Once the table was indexed, the query time was reduced to LESS than 1 millisecond.

Review long transactions. Dig into those issue to ensure efficient code is being used and that the tables and columns involved have been indexed.