SilverG33k:
Honestly, let's just catch cruise here.

Now here's a quick question, you're to build nairaland backend, how would you do that?

What technologies would you use, how would you handle scalability issues. Note than, the expected traffic for this system is 1k req/s just for fun tho

No yabbing

DC49:


In my case I would use a decoupled architecture. That is, separate the front-end from back-end and host them separately.
Front end would be an Angular app hosted on firebase, while back-end would be a Django Rest API hosted on Google App Engine with auto-scaling support to handle traffic surges.
I'd utilize a caching system like Memcached to reduce load on the back-end server, store the user generated contents like media files on a separate file storage service like Amazon S3 or Google Cloud Buckets, host the database separately on Google Cloud SQL, and finally use a CDN like Cloudflare.
This setup would be expensive though, but for a platform that gets such a huge amount of requests I'm certain the owner would be able to afford it.

DC49:


In my case I would use a decoupled architecture. That is, separate the front-end from back-end and host them separately.
Front end would be an Angular app hosted on firebase, while back-end would be a Django Rest API hosted on Google App Engine with auto-scaling support to handle traffic surges.
I'd utilize a caching system like Memcached to reduce load on the back-end server, store the user generated contents like media files on a separate file storage service like Amazon S3 or Google Cloud Buckets, host the database separately on Google Cloud SQL, and finally use a CDN like Cloudflare.
This setup would be expensive though, but for a platform that gets such a huge amount of requests I'm certain the owner would be able to afford it.

talk2hb1:
Tooh,
Waiting for Experience Bosses to Come Tutor us on this.

SilverG33k:


lmao look at this boss oh, chai

Okay, Sir... How would you handle 10 backend devs working with 4 different languages

DC49:


In my case I would use a decoupled architecture. That is, separate the front-end from back-end and host them separately.
Front end would be an Angular app hosted on firebase, while back-end would be a Django Rest API hosted on Google App Engine with auto-scaling support to handle traffic surges.
I'd utilize a caching system like Memcached to reduce load on the back-end server, store the user generated contents like media files on a separate file storage service like Amazon S3 or Google Cloud Buckets, host the database separately on Google Cloud SQL, and finally use a CDN like Cloudflare.
This setup would be expensive though, but for a platform that gets such a huge amount of requests I'm certain the owner would be able to afford it.

SilverG33k:
Honestly, let's just catch cruise here.

Now here's a quick question, you're to build nairaland backend, how would you do that?

What technologies would you use, how would you handle scalability issues. Note than, the expected traffic for this system is 1k req/s just for fun tho

No yabbing

cixak95211:


NL say she has 2.5M members, if that is anything to go by then indeed it has to be a microservice. There must be very bad guys amongst these members, so an API Gateway is a must for the backend. I'll likely go with Kubernetes so it handles upscaling and downscaling for me without having me loose sleep. NGINX is very much ok as an LB for both frontend and backend, i prefer a round-robin as against weighted since the K8 engine will spin up equally weighted infra. For the DB, i'll likely go with Mongodb since it will allow me run a search engine without the need for Algolia . . oh well Algolia can come later , but at least I can do database sharding right out of the box. You dont expect me to dump 2.5M parent records which will span into hundreds of millions, imagine one user has 100 posts, which is way likely to be more each week. Fact is i cannot dump those in one single DB, no maatter hwo powerful it is, It has to be redundant. Redis is needed for caching frontpage topics, so i dont have to serve it all the time, for each request. I could cache with NGinx too but i prefer redis. LRU is powerful.
Choice of server infra . . a simple linux image will do, ssd based and most importantly, all servers, FE, BE and DB must be encrypted at rest. I'll use a KMS for that. I'll leave one bastion host available for maintenance work and make sure all servers only communicate via a VPC.

SilverG33k:


Database sharding how did I miss that... Excellent point there, very necessary, why would the DB be encrypted tho? Or what part of the DB needs to be encrypted, please share, because I am just considering the speed and complexity

DC49:


In my case I would use a decoupled architecture. That is, separate the front-end from back-end and host them separately.
Front end would be an Angular app hosted on firebase, while back-end would be a Django Rest API hosted on Google App Engine with auto-scaling support to handle traffic surges.
I'd utilize a caching system like Memcached to reduce load on the back-end server, store the user generated contents like media files on a separate file storage service like Amazon S3 or Google Cloud Buckets, host the database separately on Google Cloud SQL, and finally use a CDN like Cloudflare.
This setup would be expensive though, but for a platform that gets such a huge amount of requests I'm certain the owner would be able to afford it.