7 Comments

Nice one! Thank you!

Expand full comment

Really interesting! And an original approach to solving spikes :)

Expand full comment
author

thank you

Expand full comment
Dec 19, 2023Liked by Neo Kim

Thank You Neo, the best part here is reference given in the post, so people willing to deep dive can go through those also.

Expand full comment
author

you're welcome

Expand full comment

Interesting solution, Neo.

I guess the serverless architecture serves well for the spikes, but I'm not sure about the price.

Thanks for sharing!

Expand full comment

I wonder how the Exchange Lambda works

IMO, the waiting room service shouldn't be responsible on determining when a new batch of users should be let to the protected zone. This being said we need some sort of integration between the protected service and the waiting room so we could update the leaking bucket somehow.

Since any integrations typically take a lot of time and effort from both sides I was hoping to find out if there is some "creative" solution that would allow adding the waiting room as a service with low to no coding effort from the service provider perspective.

Note: Yes, we could definitely allow N users go every 1/3/5 minutes, but this doesn't guarantee that the protected service will be able to withstand it, since once we let somebody in, we do not know if they already made a purchase or still in-progress.

In any case, the article is great, thank you for sharing!

Expand full comment
Error