One of the only Indian tech channel that I would say is top notch. I try to watch only non Indian tech channels though am Indian because I feel Indian youtube channels they focus more on making things look complicated as possible and using jargons over that to make the learner feel overwhelmed. But this channel is the only Indian channel I just love because every explanation is built from scratch right in front of you with easy to follow visuals, and the explanation blossoms spontaneously. You rock Vadgama bro. I have said it many times before, will say it again in future- I am so grateful for finding this channel.
Thanks for the video! I need a clarification. Bulkhead pattern and circuit breaker pattern are two different solutions to prevent the same problem (i.e. a microservice, say, an order service becomes unavailable after sometime if another service, say, payment service takes too long to respond to the requests from order service.). Is my understanding correct? If my understanding is wrong then please explain the difference between the problems prevented by the two patterns.
Does the Resilience4j manage Tomcat main ThreadPool settings based on the configuration we are providing while creating bulkhead or do we have to make sure on our own to keep the total and thread count for each bulkhead to be less than or equal to the total ThreadPool size of the Tomcat?
Just as i was watching the video and reached half way, did i realise the significance of automated unit testing for such scenarios. Most of us would design the system, as per access patterns, but how many go out of way to automate testing for them.
Quick question: If threshold is 3, and 3 threads have gone into wait state as the other service is not responding, any further requests will get a default response? But what about these 3 threads? Are they released after some time out and then how is the count maintained then? Also, I'm kinda confused how this is different from Circuit Breaker. They both are keeping track of a threshold and returning default responses till the other service is back.
First and foremost, great video series and explanations. Thanks a lot. I was looking for a good source of info on microservices and you have nailed it. Couple of questions on your video: 1. Isn't bulkhead pattern similar to the circuit breaker pattern that you mentioned in another video? Instead of number of failures, now the factor is resource usage. 2. As a design pattern, which one is generally recommended for a service: To have a threadpool in a single service or have multiple instances of a service with a single thread in each? Thanks in advance for your responses :)
1. Circuit breaker is used to help the other service return to normal while bulkhead us used to protect current service use only limited resources. 2. Ideally even with multiple instances of a service it's good to use threadpools in each instance though need to set count accordingly and if required use external service like redid for coordination
Could u plz do a video about how to calculate internet traffic? For system design purpose, one of the step is to calculate the volumnes, eg a video usually takes 1000 MB for 10 min... etc... A text how many bytes etc... Thank you!
You make amazing videos buddy. And it is very much what I need. I actually wanted to ask you on how your bring up processes in case of a fail over scenario.
Do you mean when a process stops responding to any requests? I think health-checks will help in that case (eg: by Load Balancer). All load balancers have active healthchecks which avoid requests going to failed process. Then its upto your alerting systems to know process is stuck and needs a restart. Process itself won't be able to perform any function.
@@DefogTech there will be active requests being processed. While processing these requests, suppose the server goes down. We will have to handle the fail over process and start the same processes from the failed stage in another server. Ps :These are asynchronous requests. So the client will not be waiting for an immediate response.
That state of processing will need to be saved externally so that next process can continue. Example if request is taken from queue use transaction where only if request is processed successfully the message is acknowledged
Absolutely. Though, istio is not yet integrated within many PAAS platforms. Its currently only available for K8s. With servlerless and cloud, slowly all these patterns would be available for developers to configure.
@@DefogTech Thanks for double defogging it :) Btw just watched your another video (ruclips.net/video/QiXK0B9FhO0/видео.html) that explains this nicely ! Thanks a ton .. keep up the good work sir !
The content you put up is amazing! Just out of curiosity, how do you learn all these new patterns. Do you implement these at your job etc? I assume you work as a software developer correct?
Yes, I work as software developer, but unfortunately don't get to work on any good technologies shown in the videos. I just read a lot and try to stay relevant, and have recently started to put that into videos. Hope it's helping
@@DefogTech That's awesome that you stay up to date. What resources do you use to do this? Also don't mind me asking, what tech do you use at your role?
the problem of this pattern is how do we decide the concurrent call number? It seems like we need to estimate according to all downstream microsevices,and allocate to each remote service some resource, which means this service is strong coupled with downstream service right?
true, but once you decide to call a microservice it is a dependency required to fulfil a function, so its not strong coupling because in the future the microservice being called can be replaced with another one with same API
![Jack Harlow - Hello Miss Johnson [Official Music Video]](http://i.ytimg.com/vi/Q86_nlRoIGw/mqdefault.jpg)
One of the only Indian tech channel that I would say is top notch. I try to watch only non Indian tech channels though am Indian because I feel Indian youtube channels they focus more on making things look complicated as possible and using jargons over that to make the learner feel overwhelmed.
But this channel is the only Indian channel I just love because every explanation is built from scratch right in front of you with easy to follow visuals, and the explanation blossoms spontaneously.
You rock Vadgama bro. I have said it many times before, will say it again in future- I am so grateful for finding this channel.
I see Defog Tech video, I upvote
Beautiful description of the pattern!
Pure genius! Buddy your videos are gold.
Bhai Bahut sahi samjahte ho i love the way you make us understand
Very clear explanation of a complex subject
thanks for explaining this concept with great clarity, detail and fluency
Nice one...this is much easier using the microprofile tolerance library...
Thanks for clear and concise explanation.Great video!
Very Good Explanation , thank you !!
Such clear explanation
Thanks, your videos are motivation to implement a POC for my blogs.
Thanks for the video! I need a clarification. Bulkhead pattern and circuit breaker pattern are two different solutions to prevent the same problem (i.e. a microservice, say, an order service becomes unavailable after sometime if another service, say, payment service takes too long to respond to the requests from order service.). Is my understanding correct? If my understanding is wrong then please explain the difference between the problems prevented by the two patterns.
Thanks a Million ! All videos in this channel are just wow ... Can you please make video on Concurrent HashMap internal working ...
please keep adding more videos
You got race case in the first example, you should change int to AtomicInt
Your explanation is awesome..but one thing i want your example code for every tutorial that will help us practically more .!
please add video on handling concurrent transactions in spring boot and micro services thanks
Thank you for uploading the video. Very clearly explained.
Thanks for your video!
You are awesom. I need a mentor like you
Great explanation!
Nice explaination. Can we get this file as notes?
Does the Resilience4j manage Tomcat main ThreadPool settings based on the configuration we are providing while creating bulkhead or do we have to make sure on our own to keep the total and thread count for each bulkhead to be less than or equal to the total ThreadPool size of the Tomcat?
Good information .. Keep doing good @Defog Tech , Could you please make a video on EventDriven concept in microservices.
Sure, that's a good idea. Thanks
Could you give some pointer as to how to design a distributed bulkhead config?
Can you please update the videos to the current year.
Just as i was watching the video and reached half way, did i realise the significance of automated unit testing for such scenarios. Most of us would design the system, as per access patterns, but how many go out of way to automate testing for them.
Could u please make more videos🙏
Quick question:
If threshold is 3, and 3 threads have gone into wait state as the other service is not responding, any further requests will get a default response? But what about these 3 threads? Are they released after some time out and then how is the count maintained then?
Also, I'm kinda confused how this is different from Circuit Breaker. They both are keeping track of a threshold and returning default responses till the other service is back.
Awesome explanation. Could you please make videos on design patterns
Good explanation! Thanks for posting.
First and foremost, great video series and explanations. Thanks a lot. I was looking for a good source of info on microservices and you have nailed it.
Couple of questions on your video:
1. Isn't bulkhead pattern similar to the circuit breaker pattern that you mentioned in another video? Instead of number of failures, now the factor is resource usage.
2. As a design pattern, which one is generally recommended for a service: To have a threadpool in a single service or have multiple instances of a service with a single thread in each?
Thanks in advance for your responses :)
1. Circuit breaker is used to help the other service return to normal while bulkhead us used to protect current service use only limited resources.
2. Ideally even with multiple instances of a service it's good to use threadpools in each instance though need to set count accordingly and if required use external service like redid for coordination
A very good video sir. Thanks a lot
Sir, you should critique my videos instead of thanking me. Allow me to learn and grow :)
Thanks so much for this tutorial.
wow, clear and concise .. please keep doing more videos !!
Could u plz do a video about how to calculate internet traffic?
For system design purpose, one of the step is to calculate the volumnes, eg a video usually takes 1000 MB for 10 min... etc... A text how many bytes etc...
Thank you!
Thats a good idea. Will add it to my TODO video list. Thanks!
@@DefogTech thank you 😄
Brother, Why you don't make videos now?
You make amazing videos buddy. And it is very much what I need. I actually wanted to ask you on how your bring up processes in case of a fail over scenario.
Do you mean when a process stops responding to any requests? I think health-checks will help in that case (eg: by Load Balancer). All load balancers have active healthchecks which avoid requests going to failed process. Then its upto your alerting systems to know process is stuck and needs a restart. Process itself won't be able to perform any function.
@@DefogTech there will be active requests being processed. While processing these requests, suppose the server goes down. We will have to handle the fail over process and start the same processes from the failed stage in another server. Ps :These are asynchronous requests. So the client will not be waiting for an immediate response.
That state of processing will need to be saved externally so that next process can continue. Example if request is taken from queue use transaction where only if request is processed successfully the message is acknowledged
@@DefogTech thanks that was the plan. To save it in a database. And if some servers fail, all failed processes have to be rerun. Right?
Correct.
@Defpg, great video.Is bulkhead pattern implemented in Service Mesh or API Gateway? Can you explain?
Its a general pattern that can be used in both
is there an available github repo to show this in action at all please?
Would sidecar proxy technologies like istio handle these issues better in Microservices .. instead of handling it in business logic layer ?
Absolutely. Though, istio is not yet integrated within many PAAS platforms. Its currently only available for K8s. With servlerless and cloud, slowly all these patterns would be available for developers to configure.
@@DefogTech Thanks for double defogging it :) Btw just watched your another video (ruclips.net/video/QiXK0B9FhO0/видео.html) that explains this nicely ! Thanks a ton .. keep up the good work sir !
Is this what happened with that aws us east issue last month?
The content you put up is amazing! Just out of curiosity, how do you learn all these new patterns. Do you implement these at your job etc? I assume you work as a software developer correct?
Yes, I work as software developer, but unfortunately don't get to work on any good technologies shown in the videos. I just read a lot and try to stay relevant, and have recently started to put that into videos. Hope it's helping
@@DefogTech That's awesome that you stay up to date. What resources do you use to do this? Also don't mind me asking, what tech do you use at your role?
the problem of this pattern is how do we decide the concurrent call number? It seems like we need to estimate according to all downstream microsevices,and allocate to each remote service some resource, which means this service is strong coupled with downstream service right?
true, but once you decide to call a microservice it is a dependency required to fulfil a function, so its not strong coupling because in the future the microservice being called can be replaced with another one with same API
@@DefogTech I got your point,thanks for your explain ^_^
Wait why can't we use Async calls in this case?
🙏
Can't use semaphore for this use case instead of third parties library?
yeah, sure can, internally these libraries must be using something similar.
@@DefogTech thank you so much for response.
🔥🔥🔥
Who owns this channel? Will it be active again??
I own it. It will be active again I promise :)
go semaphore =)
why do you have subtitles disabled? why why why
I haven't disabled it. The video is not popular enough for RUclips to generate one I suppose.