Clip
Summary
Two weeks ago, the team completed the Collaborate V1 database migration to its own dedicated SQL infrastructure. Previously it shared the Evo core platform's SQL cluster. Migration took one Tuesday evening, with 6 minutes of Collaborate downtime and zero impact to the Evo platform. Post-migration, core platform SQL CPU fell by 7.5% day-on-day and 13.8% adjusted for throughput vs. the previous week. That's headroom for EVO to keep scaling.
Screenshots
Transcript
So using this simple filter, you can see added people, you can see deactivated users, and you can see if the e-mail has been sent or not sent. Yeah, and yeah, that's kindly it from me. Thank you very much.
And I think Lucian is next.
Lucian Rimoti 50:28
Yes, thank you. I'm going to share my screen. All right, so two weeks ago, we completed something that's been a long time coming, which is moving Collaborate V1's database on its own dedicated SQL infrastructure.
A quick bit of history. on this. So Collaborate started life inside the aCloud code base, later workspace, then EVO. Sometime back, we pulled the application out into its own independently deployable code base, but the database stayed on our shared SQL cluster alongside the core platform.
With EVO adoption growing, Marius Pool's Dean in to basically plan and run the migration, which gives collaborate its own dedicated infrastructure and safeguard the performance and capacity of the core platform as we scale. It was completed on last Tuesday evening with support from Oli Darling. Total downtime for collaborate was 6 minutes with zero impact to Iova, which is a really good and strong result.
Dean has left me with some graphs and I want to go quickly over these. So during core UK business hours, the day after migration, CPU usage across Evo platform SQL fell by 7.5% versus the same period the previous day. That's a like for like comparison adjusted for throughput.
So expanding the comparison a bit further, the day after migration was 8.9% lower than the equivalent window a week earlier. And again, throughput adjusted. And on the last slide, you can see that looking at last Friday morning ramp up window from 8.30 to 9.30, CPU usage was 26.7% lower than the same window on Tuesday before migration. migration throughput during that Friday window was about 15% lower though.
So once we adjust for that, the like for like reduction lands at 13.8%. So to wrap it up, high single to low double digit reductions in resources, resource usage on the core platform SQL infrastructure.