Hi all, we are running many tests comparing Meteor 3 with Meteor 2, but I need help deciding the best way to share the results.
The main goal of this study is to understand the impact of Fibers on Meteor performance or, a better description, the impact of removing Fibers in Meteor 3.
These tests are full of information and have graphs in different areas, such as Event loop lag, Fibers created, Memory usage in many levels (heap spaces), CPU usage, so my question is:
What is the best way (format) to share these results with the Meteor Community? You can choose 1 or 2 options.
- YouTube Videos
- Blog Posts
- X (Twitter) Threads
- Forums Posts
- Newsletter Articles
0 voters
I imagine this work will take many weeks because we can develop different test scenarios, and we are open to ideas.
What are we already testing?
- REST calls with heavy CPU processing (Fibonacci)
- REST calls with heavy MongoDB operations ($inc ops in the loop with findOne’s)
What are we planning to test next?
- Method calls with many connected clients (Artillery + Playwright + AWS Fargate)
- Subscription calls with many connected clients as well
From there, we should branch off to more tests based on the findings.
Also, there are variations in container sizes and quantity to be tested. We did this a few weeks ago for a Workshop about Load Testing, and it was great to find the best container size for a specific workload. Unfortunately, for most of you, it was in Portuguese.
Links to code (quavedev/meteor-template on GitHub):
REST for Meteor 2
REST for Meteor 3
Spoiler about REST tests: Memory usage seems to be better in Meteor 3, and CPU usage is the same, but processing more requests per second.
I hope this study can help companies decide about migration and how to take advantage of Meteor 3 in the best way.