Network Reliability Engineering Community

August 13, 2019 Community Standup Notes

3 releases, all related: first thing is that we separated curriculum from platform releases to enable faster release cycles, esp for curriculum.

Also implemented Collections.

Still running in GCP for supporting services, but have moved to bare-metal services from Packet for primary services. Double the core speed, but much less RAM (don’t need that much, because we clean stuff up automatically about every 30 mins). The per-node price is now $.15 lower, with better performance. Lesson load times are a little faster, looking to optimize further.

Ethan: does this help with scale, eg load across the system as more users come?

Matt: Yes, looking ahead to that. Both performance and paying down technical debt while we’re still early.

Ethan: Will add the release blog to the PP weekly Infrastructure newsletter.

1 Like

Hello Icaywood,

I have one question to ask from you for on your recent post stand up notes. you said you can clean stuff up automatically in every 30mins.my question is how? can u explain me in short?.

Thankyou.

@mierdin Can you comment? You whizzed through that pretty fast.

@varsha Recordings of community meetings are also available on YouTube: https://www.youtube.com/watch?v=GnEz0j3-d7Y

This just means the platform shuts down any lesson that’s been inactive for 30 minutes so we aren’t wasting compute resources. It’s a configurable value but in production it’s 30.