tl;dr

This Friday is Black Friday, and we have concerns with one client (Headout) this year.

We also want to achieve a first migration dry run to the new content database.

Goals for This Week:

Black Friday

This will be our main focus of the week. Here is our plan to manage this week smoothly:

- Add more machines to handle the load and the expected increase in load from all of our users.

- Isolate computing resources for Headout so that it won’t impact other clients if something goes wrong.

More about Headout isolation.

We have a pool of lambdas that we can use simultaneously to compute content on the public content API. Today, we have 10k lambdas available. Headout usage brings our infra to use all our lambdas simultaneously, then have an impact on other clients. We want to isolate them into another AWS account and have access to another pool of lambdas.

- We don’t anticipate a huge increase in cost because we never reached 10k lambda without Headout.

- We will limit the maximum concurrent number to 2k.

- We already have this account set up for shadowing on the Public Content API. This will have an impact if we need to do shadowing on the Public Content API -> No scheduled shadowing for now.

In the long run, we still need to come up with a solution to avoid isolating Headout.

Database migration

On the front of the scaling, we expect to finish the dry run and have it run on most of the repos.