How Tinder / Bumble does Matchmaking? System Design Charcha with  

Thank you guys. That is what I looking for without preparing/remembering. Just two master are talking each other and solving the problem step by step.

Appriciate you Keerti maam for making such informative content love from Bangalore, I do share your videos in our groups

I'd strongly recommend discussing less about the rationale behind certain requirements like why the active filter is important and more about the infrastructure design.

Really amazing series. We want such rare discussions!!!

Wow...thanks. Its a great video which illustrate how we should approqch a system design interview 🙂

We can put Rate limiter in this system as well because might be if someone using hack script and able to swipe 100 profiles in 1 min then we have to stop it.

hi Guys, one doubt... I mean this might be beyond scope of this video but.. What if user changes his location from Mumbai to Banglore and while he was in Mumbai, we had some recommendations in Cache. And next day when he opened the app in Banglore, all recommendations need to be updated... So again we might need to go through the DB, update the cache of recommendations. OR Until user changes his "location preference", we will not re-run queries and update the cache. But if we do so, user will still see recommendations from Mumbai, even if he is in Banglore. or maybe we can just give the user a notification saying, your location seems to be changed. We strongly recommend you to change location preference as well. Can you please share your opinion on this?

I have been loving programming. And should me go to this channel. But now I love you ❤.

Not sure about Cassandra, I feel when we have a very complex search query structure like - filter on age, location, interests, gender, hobbies etc a search engine that supports text based search (basically quick indexing on props) like ES may be an option. But Cassandra looks like a horrible choice for a read-heavy system like Tinder :( Surprised to see Gaurav was okay with this lol

22:50 regarding the choice of using C*, yes it compresses the data (column wise). But, match making is read heavy operation. So, C* vs SQL, there is nothing great that C offers. The biggest advantage is horizontal scaling when it comes to storing data. If you store entire Country data in a single Instance, then it makes sense. Or If you store each city data in a separate Shard, then Postgresql is also good enough; it supports geo spatial indexes & queries. If each city is a shard, then we're thinking of storing 2-3cr (20-30M) in a single table. It's something that sql reads can easily manage. So, it boils down to companies expertise in db: sql vs nosql. Using sql makes it simple & easy to manage. Managing C is a pain and has a huge maintenance overhead. Eventually, they end up managing lots of DBs, say, 100k shards/DB instances. So, need to see if C can scale to have these many shards or can make it easy to manage with lesser no.of shards.

Thank you soo much broo for sharing ❤❤

Thank you so much keerti ma'am & Gaurav bhaii... Please make video on system design of ChatGpt if possible.. Thank you

he is really great..! Thanks for sharing this video

2mc2 would be around 2 trillion instead of 4. You'll only match two people

100 is totally fine 🤣🤣🤣🤣🤣

hello, thank you for your videos they are really useful. Are you plannig to complete software design patterns series?I guess there are only 7 videos🥲?