I only knew we can scale nodeJs server, but have never tried to do so. Thanks to you for making me(us) know scaling server in nodeJs. This is worth a subscribe.❤️☘️
A nice solution that you can use for small projects. However, if you are looking for a solution for more complex projects, instead of this, you should run multiple node instances and put a load balancer in front of them.
@@StLouis-yg8jp Because they would share the same memory on your EC2 (I mean the machine you run application), it's a drawback for ram intensive tasks.
Great explanation sir , I am watching your videos from last year now you are doing too good . I am following you from my 3 different id's. Great work keep going on .
I deploy to Heroku using a Hobby dyno and I have noticed my node project barely uses any CPU or RAM when stress testing. So i added more workers, because its only 1 cpu I had to hardcode a worker amount - i found 6 was good for my app. My requests per second and response times improved drastically - 6 fold you could say. I will now try cluster to see what happens but i think its the same thing... although its recommended to not do this and let Node handle the load, i cant help but notice adding workers and clusters utilizes more of the Servers resources and speeds up the response per second.
Just a reminder: Use worker_threads if you can. A thread will also run in parallel on a CPU with multiple cores (*almost, you just hope that the CPU distributes the workload). A thread has it's own context, event loop, and can work on a shared buffer with other threads. A process would spin up an entirely new Nodejs runtime, event loop, context, with no ability to work on shared memory. They have their differences.But in both cases most javascript data types have to be copied around, and in both cases if the CPU has only one core with one logical processor then it can't do parallelism and instead it does system level async stuff. And yes different processes crash independently, but worker threads can also crash independently, you just have to handle the errors.
This is good for vertical scaling and only if your machine has more than one CPU thread. If your CPU is single threaded one can also look into kubernetes. After understanding kubernetes pods you can see how beautifully they solved this problem on an infra level. Also great video bro!, keep em coming....
Hey brother...i like the way you explain things...one doubt in this lecture is that how to apply this cluster module when we are using controllers and routes style of architecture...Hope you will reply my question🤞
Which is better scaling with Cluster create instance upto n number of logiacal core Or using Docker to create multiple container and using nginx as reverse proxy ?
You can use PM2 to run your application in Cluster mode and also it servers as your process manager ie if your application by default crashes then it will be restarted automatically
@@nishantbhavasar5047 using pm2 will start it in cluster mode also. You possibly will have to use load balancer and use pm2 to create different worker threads.
Just got this on my recommendation and I have a question, why kill the worker and fork another after each request ? Is there a performance benefit and what about the performance penalty in fork+kill code execution ?
i suppose to deploy my site in aws, should i need this node cluster module or aws load balancing is good?please share your thoughts which one is better
os.cpus().length will return the number of cores not threads (I beleive). So I have a few doubts 1. If I have 2 cores 4 threads cpu then will it be on full potential this way because it looks like it will create 2 proces only. 2. If I have to host say 3 apps on a 2 threads processor so how should it done. Should I create 2 processes for all 3 apps resulting in 6 total proces?
1. yes os.cpus().length gives u the number of cores, if I said something else its slip of my tounge. 2. You should only have the number of instances of you app === number of cores of a CPU 3. If you have a single core CPU then you should run 2 instances, so as to have Zero downtime, in the case one of the instances go down. Note: Regarding ur query about cores and threads, you can read this link and will provide you more insight about that and your question will be answered. www.quora.com/What-does-2-core-4-threads-mean-in-a-cpu-I-am-talking-about-this-cpu-ARK-Intel%C2%AE-Core%E2%84%A2-i5-520M-Processor-3M-Cache-2-40-GHz#:~:text=It%20means%20that%20the%20CPU,pipeline%20to%20process%20another%20thread.
@@mafiacodes By this link you mean ... ? PM2 cluster mode with command pm2 start app.js -i max creates 8 apps on my 4 core 8 threads processor and according to you 4 is best. It got me little confused here :/ and also Is your name really Truly (just asking) :|
Bro I can’t really come up with an answer since I have an old quad core i5 4670k, and pm2 only does 4 coz of 4 threads. In ur system what do u get for os.cpus.length? Coz what I did was in documentation. And yes my name is Truly.
Ok solved, it’s because since it’s hyper threading. U have a 4 core cpu with 8 threads, that means that for each core the os provided u with 2 virtual cores and therefore ur getting 8, for more info read wiki: en.m.wikipedia.org/wiki/Hyper-threading In the end a gr8 question and I think it will help out others also. Solved 💯
Hello, something happened to me different than video, I do not know why only same process response all the time for every requests.(Before adding zero downtime property) Do I miss something? Great video btw!
It is because in node v4.x and newer versions round-robin scheduling is disabled by default on Windows for performance reasons.Simply just add this code " cluster.schedulingPolicy = cluster.SCHED_RR; " if someone has same problem.
I am getting the same results using 1 process as with 8. Both tests are finished in ~0.43 seconds and output ~2290 requests per second. Could this be due to it being such a simple program that it doesn't make a difference that I have multiple processes running on it? I have an Intel I5 8600 processor with 6 cores and ~4000mHz.
Sir , how to do if we have routes , like I mean in my project I have a routes folder in which multiple JS files are there and each of it contains GET request, in such case how to perform this?
thx for the tuto my friend, although i wanna ask what is benefits for killing the worker and fork a new cluster? why cant we just leave the workers without killing it eact time its called.
cluster module vs child process i am confused between them, cluster module uses cores of cpu while child processes? could you make a difference video? please
Where were you a year ago when I first heard of this stuff and went on to figure it out on my own 😢
Subscribe like share
Best RUclips for NodeJS important topics.
this is one of the awesome video on node cluter module on RUclips yet...
You are helping to do my app on production.
Please make more videos like this so that we can learn real-life stuff.
this video is awesome. but i stucked as you didn't tell where to use this loadtester?
This is just what I was looking for thanks man. Please post more core concepts like these. Great Work
Same here too, I've been looking for this type of example didn't see any, what I found is how they work not how to use them.
This is so cool, your explanation was superb, didn't knew a single thing about cluster, but was able to setup the whole thing with so much ease.
I only knew we can scale nodeJs server, but have never tried to do so.
Thanks to you for making me(us) know scaling server in nodeJs. This is worth a subscribe.❤️☘️
Awesome 🔥
As mentioned in the description, every line was explained in a detailed manner.
Amazing real world project tutorial.
A nice solution that you can use for small projects. However, if you are looking for a solution for more complex projects, instead of this, you should run multiple node instances and put a load balancer in front of them.
Yes
Pls how can I do that, is there any tutorial source you can provide to learn this 🙏, thanks
Why not have multiple node instances all using the cluster module? Wouldn't that be the most efficient?
@@StLouis-yg8jp Because they would share the same memory on your EC2 (I mean the machine you run application), it's a drawback for ram intensive tasks.
that 's the reason i subscribed to your channel , thanks a lot bro thanku thanku thanku
Thanks man for sharing important information!
very nice explain clustering
Thanks for clear explanation
Beautiful. Thank you for sharing 🌷
Great explanation sir , I am watching your videos from last year now you are doing too good . I am following you from my 3 different id's. Great work keep going on .
Great explanation bhai 👍❤️🔥
I would suggest installing nodemon as dev dependency or global dependency
Great work 👏
Thanks ✌️
Excellent example
subscribe like share
Subscribed awesome explanation.
Amazing video , thanks bro
Thank you man, you saved the day.
Glad to hear it
I deploy to Heroku using a Hobby dyno and I have noticed my node project barely uses any CPU or RAM when stress testing. So i added more workers, because its only 1 cpu I had to hardcode a worker amount - i found 6 was good for my app.
My requests per second and response times improved drastically - 6 fold you could say.
I will now try cluster to see what happens but i think its the same thing... although its recommended to not do this and let Node handle the load, i cant help but notice adding workers and clusters utilizes more of the Servers resources and speeds up the response per second.
Interesting
outstanding explanation bro. Keep posting such quality content 🔥🔥
Your fluency is awsome, just make your speed a bit slow - it will be great
this was the great video sir 😎😎 i use compresion with cluster module and it give realy great result thans sir to make this awesome video. 👍🤩😍.
So good. It's a very good way to scale a simple application for starters.
Just a reminder:
Use worker_threads if you can.
A thread will also run in parallel on a CPU with multiple cores (*almost, you just hope that the CPU distributes the workload).
A thread has it's own context, event loop, and can work on a shared buffer with other threads.
A process would spin up an entirely new Nodejs runtime, event loop, context, with no ability to work on shared memory.
They have their differences.But in both cases most javascript data types have to be copied around, and in both cases if the CPU has only one core with one logical processor then it can't do parallelism and instead it does system level async stuff. And yes different processes crash independently, but worker threads can also crash independently, you just have to handle the errors.
very very great work
very clean explanation, but wondering why we need to restart a worker after each request ?
No need
Amazing tutorial, thank you
Glad you enjoyed it!
This is good for vertical scaling and only if your machine has more than one CPU thread.
If your CPU is single threaded one can also look into kubernetes. After understanding kubernetes pods you can see how beautifully they solved this problem on an infra level.
Also great video bro!, keep em coming....
how kubernetes solves this in infra level?
Hey brother...i like the way you explain things...one doubt in this lecture is that how to apply this cluster module when we are using controllers and routes style of architecture...Hope you will reply my question🤞
@Not Top Voices thanks man i got it
Good explanation
Nice tutorials, helping me alot. Keep uploading
Subscribe like share
Good job bro. What is your keyboard name ?
wow, such a great explanation 😃
I wish I could have a love reaction button.
Very useful information.
great explanation.
Should we need to do load balancing after this also?
great one
PHENOMENAL VIDEO BROTHER!! Mashallah!
Which is better scaling with Cluster create instance upto n number of logiacal core Or using Docker to create multiple container and using nginx as reverse proxy ?
Docker with nginx as reverse proxy
Super awesome video sir 💜❤💙
Subscribe like share
Nice tutorial
Does clusters are still useful in recent Nodejs versions or is there any alternative of it?
Thanks for this awesome video 👍🏻
Although the content here is superb, something else caught my eye! What did you have to install to have your comments use a different font?
Clear explanation thanks
You can use PM2 to run your application in Cluster mode and also it servers as your process manager ie if your application by default crashes then it will be restarted automatically
yes or it has inbuilt load balancer also so it's a good approach then clustor module
@@nishantbhavasar5047 using pm2 will start it in cluster mode also. You possibly will have to use load balancer and use pm2 to create different worker threads.
Good explanation,
please try to Make videoes on nodejs interview questions
Thanks
Hi truly, could you please make a tutorial on how to use Typescript in Node projects .
P.S You are a great Teacher.
Well I guess I will try sometime
Superb 👐🏻👏🏻
Thank you man. I love you!
Great video.! Just a quick question the performance will be the same if I just run in the vps pm2 start -i max ????
Just that if u have 1 cpu only one instance and if cpu>1 then instances equal to cpus
Create content like this you will get more views
Thanks for the amazong video but in my case I am getting the response from the same pid always, there is no round-robin happening
Hey what are you using to test the performance of your node application
Interesting... Thank you
This is amazing
Just got this on my recommendation and I have a question, why kill the worker and fork another after each request ? Is there a performance benefit and what about the performance penalty in fork+kill code execution ?
What kind of terminal ur using I tried multiple time in cmd but it's not working
Amazing Bro!!
i am able to fork only one worker despite having 4 cores. Any idea why?
How do you deal with cron job(s)? They get forked as well to nodes?
great video
Please make video nodejs microservices
how to get those benchmarks
For what kind of applications clustering API is ideal choice
@mafiacodes If we do this, wouldn't it increase the load on our database and our DB may crash, how should we handle that scenario?
Amazing video but can we use this same idea on real time project like i already have a project so can i use this for fast performance in my project
Yes
what is the difference between forking and creating a new worker thread ??
so how do i use this with express code
it is really a super useful video
Please anyone tell me how to get that log test results
Could you please make it using pm2 package?
Truly Amazing :)!!!!
you are awesome
Thanks. Which theme and font you're using?
Why do I have to kill the process?
i suppose to deploy my site in aws, should i need this node cluster module or aws load balancing is good?please share your thoughts which one is better
If u wanna pay use aws else you can do it urself like this or even with nginx
such a hard time with nextjs
Hello, are you using a special font for coding? if yes which is the name?
I really like it.
Thank you
operator mono
can you please tell your vs code theme and font family you are using ?
Snazzy operator and Operator mono
Subscribe like share
os.cpus().length will return the number of cores not threads (I beleive). So I have a few doubts
1. If I have 2 cores 4 threads cpu then will it be on full potential this way because it looks like it will create 2 proces only.
2. If I have to host say 3 apps on a 2 threads processor so how should it done. Should I create 2 processes for all 3 apps resulting in 6 total proces?
1. yes os.cpus().length gives u the number of cores, if I said something else its slip of my tounge.
2. You should only have the number of instances of you app === number of cores of a CPU
3. If you have a single core CPU then you should run 2 instances, so as to have Zero downtime, in the case one of the instances go down.
Note: Regarding ur query about cores and threads, you can read this link and will provide you more insight about that and your question will be answered. www.quora.com/What-does-2-core-4-threads-mean-in-a-cpu-I-am-talking-about-this-cpu-ARK-Intel%C2%AE-Core%E2%84%A2-i5-520M-Processor-3M-Cache-2-40-GHz#:~:text=It%20means%20that%20the%20CPU,pipeline%20to%20process%20another%20thread.
@@mafiacodes By this link you mean ... ?
PM2 cluster mode with command pm2 start app.js -i max creates 8 apps on my 4 core 8 threads processor and according to you 4 is best. It got me little confused here :/
and also
Is your name really Truly (just asking) :|
Bro I can’t really come up with an answer since I have an old quad core i5 4670k, and pm2 only does 4 coz of 4 threads. In ur system what do u get for os.cpus.length? Coz what I did was in documentation. And yes my name is Truly.
@@mafiacodes OK. Now I am super duper confused on my i5 8250u (4 cores 8 threads) os.cpus().length gives 8.
Ok solved, it’s because since it’s hyper threading. U have a 4 core cpu with 8 threads, that means that for each core the os provided u with 2 virtual cores and therefore ur getting 8, for more info read wiki: en.m.wikipedia.org/wiki/Hyper-threading
In the end a gr8 question and I think it will help out others also.
Solved 💯
Hello, something happened to me different than video, I do not know why only same process response all the time for every requests.(Before adding zero downtime property) Do I miss something? Great video btw!
It is because in node v4.x and newer versions round-robin scheduling is disabled by default on Windows for performance reasons.Simply just add this code " cluster.schedulingPolicy = cluster.SCHED_RR; " if someone has same problem.
@@Omer-fr1ze thanks...it solved my issue
Should i use pm2 or node cluster? Are they same?
Pm2 internally uses node cluster
I am getting the same results using 1 process as with 8. Both tests are finished in ~0.43 seconds and output ~2290 requests per second. Could this be due to it being such a simple program that it doesn't make a difference that I have multiple processes running on it? I have an Intel I5 8600 processor with 6 cores and ~4000mHz.
Can u share your code
At 0:45 Result : how to identify the result kindly guide , thanks in advance
amazing !!!!!
Just asking , will these make our app statefull ???
All instances have their own state, if u want to use single state for all of them though not necessary use Redis pub sub.
I have a lot of endpoints, do i have to kill the process on every one of them?
This is an example only, you don’t need to kill any process
@@mafiacodes Omg you replied. Thank you so much, great video as always!
Is it advisable to kill the worker in every process? In my case in every router.
Nooo, that’s only for demonstration
@@mafiacodes Okay, thank you :D
@@mafiacodes I'm curious about how to use the cluster in a socket io application? Is it a good idea to use both of them together?
Yes of course
Sir , how to do if we have routes , like I mean in my project I have a routes folder in which multiple JS files are there and each of it contains GET request, in such case how to perform this?
I am also looking for the same brother, did you get any information on this ? Is yes, could you please share some resources to study please ?
thx for the tuto my friend, although i wanna ask what is benefits for killing the worker and fork a new cluster? why cant we just leave the workers without killing it eact time its called.
Killing workers is only for demonstration, and you don’t have to do it manually, only if it fails for some reason then a new worker will be spawned.
@@mafiacodes thats what i thought. thx
What is race condition in which cases it occurs, I have heard that using cluster module may cause for race condition.
I don’t think so, because it uses round robin approach
@@mafiacodes Thanks
cluster module vs child process
i am confused between them,
cluster module uses cores of cpu
while child processes?
could you make a difference video? please
i am new to this but I think cluster is just for making copies of the current app, but child processes could run a different app