Unleashing MAAS performance with metrics, improvements, and user insights
Dive into the exhilarating world of MAAS performance, where we continuously tweak, measure, and optimise to serve you better. This guide unravels how we measure performance, highlights recent enhancements, and even shows you how to become a part of our journey by measuring your own metrics.
We've supercharged the MAAS API performance by stress-testing it under simulated conditions. Picture this: a landscape featuring five rack controllers, 48 machines per fabric, a quintet of VMs per LXD host, and machines boasting a diverse range of features—different architectures, a half-dozen types of disks, and five network interfaces. Talk about a challenging playground!
We employ continuous performance monitoring to capture this complexity. Here's a snapshot of how it looks:
Each day, we simulate machine counts of 10, 100, and 1000, providing us with a nuanced view of how well our improvements scale. The Jenkins tool puts the REST and WebSocket APIs through their paces, capturing the outcomes in a database that fuels our dynamic dashboard:
For extra depth, we always juxtapose stable releases with their developmental counterparts, allowing us to spot and squash bugs before they proliferate. Our profiling heatmaps reveal any code bottlenecks, making them ripe for optimisation. For instance, MAAS 3.2 machine listings load a whopping 32% faster compared to MAAS 3.1, based on our test datasets.
Our performance optimisation isn't a one-off effort; it's an ongoing crusade. Here are some milestones:
Note: This only scratches the surface of our systematic performance overhauls.
Want to contribute? Capture your own MAAS metrics and share them with us. We're eager to learn about your machine counts, network sizes, and performance observations. Let your voice be heard on our Discourse performance forum.
The MAAS 3.2 development sprint has already clocked a 32% speed increase in machine listing through the REST API. But rest assured, we're not stopping there.
We're now channelling our energies towards further optimising other facets of MAAS, like search functionalities.
Ready to join us on this exciting performance journey? We'd love to hear your feedback and insights.