I'm trying to build a realtime (private) chat between users of a video game with 25K+ concurrent connections. We currently run 32 nodes where users can connect through a load balancer. The problem I'm trying to solve is how to route messages to each user?
Currently, we are using socket.io & socket.io-redis, where each websocket joins a room with its user ID, and we emit each message they should receive to that room. The problem with this design is that we are reaching the limits of Redis Pubsub, and Socket.io which doesn't scale well (socket.io emit messages to all nodes which check if the user is connected, this is not viable).
Our current stack is composed of Postgres, Redis & RabbitMQ. I have been thinking about this problem a lot and have come up with 3 different solutions :
Route all messages with RabbitMQ. When a user connects, we create an exchange with type fanout with the user ID and a queue per websocket connection (we have to handle multiple connections per user). When we want to emit to that user, we simply publish to that exchange. The problem with that approach is that we have to create a lot of queues, and I heard that this may not be very efficient.
Create a queue for each node in RabbitMQ. When a user connects, we save the node & socket ID in a Redis Set, so that when we have to send a message to that specific user, we first get the list of nodes, emit to each node queue, which then handle routing to specific client in the app. The problems with that approach is that in the case of a node failure, we may store that a user is connected when this is not the case. To fix that, we would need to expire the users's Redis entry but this is not a perfect fix. Also, if we later want to implement group chat, it would mean we have to send duplicates messages in Rabbit, this is not ideal.
Go all in with Firebase Cloud Messaging. We have a mobile app, and we plan to use it for push notifications when the user isn't connected, but would it be a good fit even if the user is connected?
What do you think is the best fit for our use case? Do you have any other idea?
I found a better solution : create a binding for each user but using only one queue on each node, then we route each messages to each user.
Related
I assigned myself with the task of implementing the chat app (1:1) for my curriculum.Among the various options I used SSE for real time chats.From the example projects I am able to implement the non persistent chat between two clients.In every examples they uses js object and array to store the res object and by iterating them they sent events to particular user.But when implementing the real time chat app the users may increase dramatically So it is not good to exhaust server resources.
I found the some of the other ways to achieve same
functionality but not sure about the performance
SSE+setInterval
I used redis Queue to push offline messages to the user.
when the user establishes the connection push all the unread chats to client.
This process happens immediately when client establishes connection with server.
I faced some problem here, as I have no way of triggering the messages in real time(when both users online).
So I used setInterval with time interval of 1 second for real time communication and write a callback function to check if the Queue is empty else pop message from Queue and sent to user as an event.
Will the above solutions affect performance ? Because I am calling the function for each connected user x 1 second interval.
Long polling
In long polling how can I find if there is new message for user and complete the request ?
Still here setInterval should be used in server side but what about performance?
Websockets
In websockets we have an unique id to find the client in the pool of clients, so we can forward message to particular user when event occurs.
Still websockets uses some ping pong mechanism to make connection persistent but resource utilization is very small as they are network calls with comparatively small data and handled asynchronously so no wastage in server resource.
Questions
How to trigger res.write only when the new message arrives to particular user?
Does SSE+setInterval or longpolling+setInterval degrades performance when user increases?
Else is there any design pattern to achieve this functionality?
Simply use websocket.
It's fast, convinient and simple.
To send message in realtime when both users are logged, find second user by id in users Array or Map and send received message to his websocket.
If you have buffered messages for disconnected user (in memory/database/redis) check it when user connects and send if it exists.
I want to make an app which lets users comment and send messages. However, the notifications for these events will have to come instantly, just like any other social-media or chat application. This is what I'm thinking of:
Web-frontend: Angular, mobile: Ioinc with Angular
Backend: Node, Mongo
Now, this is how I was thinking I'd implement real-time notification.
There's a constant socket connection between the frontend (web & mobile-app) and the backend.
Whenever a message arrives, targeted to a specific user, I'll use some kind of a Mongo-hook to send the notification to the frontend via the socket connection.
Now, the confusion with this approach is:
Would millions of socket connections work at scale, at all? If not, what is the way to implement this pub-sub kind of system? I need to do it from scratch, not using Firebase.
What if a user is offline when he receives the message in the backend? If the socket is not on, how would he get the message? Is there a way to do it using Kafka? Please explain if you have some ideas on this.
Is this the correct approach? If not, can you suggest what would be appropriate?
Would millions of socket connections work at scale, at all? If not, what is the way to implement this pub-sub kind of system? I need to do it from scratch, not using Firebase.
Yes, it can work at scale just you have to made an architecture like that. You might find this useful
Scalable architecture for socket.io
https://socket.io/docs/v3/using-multiple-nodes/
What if a user is offline when he receives the message in the backend? If the socket is not on, how would he get the message?
If he the socket is not on or user is offline, then client Socket will be disconnected. At this point, notification will not be received and whenever the user comes online you'll have make an API call to get the notifications and connect again to the socket for further operations.
Is there a way to do it using Kafka?
Yes, you can also do it with Kafka. You'll need Consumer API(Subscriber) and Producer API(Publisher)
https://kafka.apache.org/documentation/#api
https://www.npmjs.com/package/kafka-node
Sending Apache Kafka data on web page
What do you use Apache Kafka for?
Real time notification with Kafka and NodeJS
I am creating an chat application where I have a rest API and a socket.io server, What I want to do is the user will send messages to rest API, The api will persist those messages in database and then send these messages to the rabbimq queue then rabbitmq will send these messages to socket.io if the receiving user is online, Else the message will be stored in the queue and when the user will come online the user will retrieve these messages from the queue however I want to implement this in a way like whatsapp, The messages will be for a particular user and the user will only receive those messages which are meant for them i.e I don't want to broadcast messages I want only particular user to receive those messages
Chat should be a near-real-time application, there are multiple ways of modeling such a thing. First of all, you can use HTTP pooling, HTTP long pooling but some time ago there where introduced the new application-level protocol - web socket. It would be good for you, along with Stomp messages. Here you can check a brief example. And sending messages to specific users is also supported out-of-the-box example
1
To send messages to specific sockets you can use rooms: io.to(room).emit(msg). Every socket is a part of a room with the same name as the socket id.
2
I wouldn't wait for the message to be written to the database before sending it out through socket.io, your API can do both at once. When a user connects they can retrieve their messages from the database then listen for new ones on their socket.
So i currently have a chat system running NodeJS that passes messages via rabbit and each connected user has their own unique queue that subscribed and only listening to messages (for only them). The backend can also use this chat pipeline to communicate other system messages like notifications/friend requests and other user event driven information.
Currently the backend would have to loop and publish each message 1 by 1 per user even if the payload of the message is the same for let's say 1000 users. I would like to get away from that and be able to send the same message to multiple different users but not EVERY user who's connected.
(example : notifying certain users their friend has come online).
I considered implementing a rabbit queue system where all messages are pooled into the same queue and instead of rabbit sending all user queues node takes these messages and emit's the message to the appropriate user via socket connections (to whoever is online).
Proposed - infrastructure
This way the backend does not need to loop for 100s and 1000s of users and can send a single payload containing all users this message should go to. I do plan to cluster the nodejs servers together.
I was also wondering since ive never done this in a production environment, will i need to track each socketID.
Potential pitfalls i've identified so far:
slower since 1000s of messages can pile up in a single queue.
manually storing socket IDs to manually trasmit to users.
offloading routing to NodeJS instead of RabbitMQ
Has anyone done anything like this before? If so, what are your recommendations. Is it better to scale with user unique queues, or pool all grouped messages for all users into smaller (but larger pools) of queues.
as a general rule, queue-per-user is an anti-pattern. there are some valid uses of this, but i've never seen it be a good idea for a chat app (in spite of all the demos that use this example)
RabbitMQ can be a great tool for facilitating the delivery of messages between systems, but it shouldn't be used to push messages to users.
I considered implementing a rabbit queue system where all messages are pooled into the same queue and instead of rabbit sending all user queues node takes these messages and emit's the message to the appropriate user via socket connections (to whoever is online).
this is heading down the right direction, but you have to remember that RabbitMQ is not a database (see previous link, again).
you can't randomly seek specific messages that are sitting in the queue and then leave them there. they are first in, first out.
in a chat app, i would have rabbitmq handling the message delivery between your systems, but not involved in delivery to the user.
your thoughts on using web sockets are going to be the direction you want to head for this. either that, or Server Sent Events.
if you need persistence of messages (history, search, last-viewed location, etc) then use a database for that. keep a timestamp or other marker of where the user left off, and push messages to them starting at that spot.
you're concerns about tracking sockets for the users are definitely something to think about.
if you have multiple instances of your node server running sockets with different users connected, you'll need a way to know which users are connected to which node server.
this may be a good use case for rabbitmq - but not in a queue-per-user manner. rather, in a binding-per-user. you could have each node server create a queue to receive messages from the exchange where messages are published. the node server would then create a binding between the exchange and queue based on the user id that is logged in to that particular node server
this could lead to an overwhelming number of bindings in rmq, though.
you may need a more intelligent method of tracking which server has which users connected, or just ignore that entirely and broadcast every message to every node server. in that case, each server would publish an event through the websocket based on the who the message should be delivered to.
if you're using a smart enough websocket library, it will only send the message to the people that need it. socket.io did this, i know, and i'm sure other websocket libraries are smart like this, as well.
...
I probably haven't given you a concrete answer to your situation, and I'm sure you have a lot more context to consider. hopefully this will get you down the right path, though.
I am implementing a TCP chat server using node js and redis, however i dont seem to be able to persist chat data on redis using Publish and Subscribe, and hence when i have left the chat room and reentered, i will not be updated on the newest messages, how should i implement something like this?
Publish is not meant to be stored in Redis, even if you chose the disk storage. When it recieves message, it just finds the connections with requested channels and forwards to each. So, it is not storing anything. Even if it did, It should continously try to forward messages (because it's a pub/sub model) which is not very effective. Instead, you should also push (by lpush the messages to a queue, so they can be stored. And when a client connects and has no messages, it can retrieve those messages from queue (without popping, so other newcomers can also use) and then subscribe to channel and recieve new messages.
By default, redis is in memory only. You have to enable persistence explicitly.
There are multiple options, AOF every query being the safest, but probably the slowest.
More details here: http://redis.io/topics/persistence