From the course: Testing Python Data Science Code

Unlock this course with a free trial

Join today to access over 25,500 courses taught by industry experts.

Choosing speed and memory metrics

Choosing speed and memory metrics - Python Tutorial

From the course: Testing Python Data Science Code

Start my 1-month free trial Buy for my team

Choosing speed and memory metrics

“

- [Instructor] One of the things that always surprises me is how hard it is to get performance requirements from business or product people. The most common answer I get is, "Make it as fast as you can." My usual answer is, "Okay, I'll get the hardware team on it. Give me a year or two." Joke aside, you need to get some hard numbers such as you have 70 milliseconds to process a message or the application shouldn't consume more than half a gigabyte of memory. Sometimes you will need to come up with the numbers and then ask, "Is an hour to process the daily report good enough?" Once you have the performance requirements, you also need to get the SLA, which means service-level agreements. For example, 99% of the requests should return under the allocated time. Make sure to convey that adding nines is going to consume a lot of effort. If you're not sure about what are realistic numbers, I recommend heading over to Napkin…

Contents