#community-help

Typesense Usage for Large-Scale Messaging Apps

TLDR Juri questioned the viability of using Typesense in a large-scale messaging app due to data volume concerns. Jason clarified that usage depends on cost and scale considerations. They also discussed client-side search and data encryption in apps like Whatsapp and Slack.

Powered by Struct AI

2

1

5
31mo
Solved
Join the chat
May 11, 2021 (31 months ago)
Juri
Photo of md5-756d5da34cc5127c88730a39db749024
Juri
09:21 PM
One interesting question: If a make an app like discord or slack or whatsapp, where every day 100GB of messages are being sent, I shouldn't use typesense for that use case, or? I should rather have the chats cached on the client mobile Phone and then do the search on his mobile phone (that's the way WhatsApp is doing it), right?
Jason
Photo of md5-8813087cccc512313602b6d9f9ece19f
Jason
09:23 PM
I heard Slack actually uses Algolia for at least a portion of their search! They probably have several 128GB RAM Algolia clusters they use.

So you can definitely use Typesense as well for use cases like that. It comes down to a question of cost and scale, since you'd have to choose clusters to hold all your data in RAM.

1

09:24
Jason
09:24 PM
Whatsapp I believe uses client-side search only because conversations are end-to-end encrypted (I think, I hope)!

1

Juri
Photo of md5-756d5da34cc5127c88730a39db749024
Juri
09:45 PM
Time for slack to switch to Typesense too 🔥 It's since whatsapp exists client side search for costs reasons (since till Facebook aquired them they had only a few $$ of income through the subscription. Now Zucc makes big cash $$$$$$$ with analysing and selling our chat data, but they still use client side search because they don't want to lose any $ that would make the stock holders sad). Their end-to-end encryption today is still a Joke anyone can break it
Jason
Photo of md5-8813087cccc512313602b6d9f9ece19f
Jason
10:00 PM
Hahaha!

1