What is Event Driven Architecture (EDA)?

preview_player
Показать описание

What is an event-driven architecture, or EDA? And how does it enable real-time user interactions, pluggable microservices, and extensible streaming and data analytics?

In this lightboard video, Whitney Lee from IBM Cloud, visually breaks down the answers to these questions and many more, as well as explains the several advantages and opportunities that an event driven architecture provides for developers and organizations in comparison to a request/response application architecture.

#EDA #FaaS #Serverless
Рекомендации по теме
Комментарии
Автор

IBM you're so good at explaining concepts with these videos. I remember watching one on the difference between an API and an SDK some time back and it was very helpful. I now understand that the event backbone is the most important part of the system. I guess other services can subscribe to it(either to all its events or just to some events that it receives) and process the event data however they want. Thanks for an awesome video

thecodeninjaeu
Автор

The presenter shared a few use cases. I like the presenter to finish the check-out use case first before jumping on the music streaming. How do you handle distributed transaction of the check-out flow in the event-driven architecture? If the user cancels his transaction, for example, there should be a compensating message. How about duplicate events published? Idempotent consumers are required to de-duplicate the message. I think EDA is hard and complicated, it's not what the presenter said it's so easier. You mentioned Kubernetes at the beginning, what is it to do with EDA? Eventual consistency shall be discussed when using EDA. etc. Thank you for the knowledge sharing.

mikedqin
Автор

Very nice explanations. I am very impressed that she writes through a glass so that we can read it ...

Demran
Автор

This channel should feature on Netflix

heyamjoe
Автор

Looking at the First 20 Seconds and asking myself: What is with that lightning setup?

TheOnlyEpsilonAlpha
Автор

That's a lot of code waiting for app and API developers to write. That's a lot of logging to sanitize and onboard to the backbone. That's a lot of data streams for admins to understand and write actionable if - then statements. That's a lot of reporting for owners and perf engis to sift through. There's a lot of IT professionals working hard to ensure this kind of infrastructure for multiple sources of data and systems scale appropriately. Hire good people to run good systems and you're winning.

jlhjedi
Автор

Love this series. Understanding so much in so little time about fundamental concepts.

Simon-feti
Автор

One of the best, clear and concise introduction i found.

vimalneha
Автор

these videos are so good to get into hard topics

daniloespinozapino
Автор

First of all, great video. I like the prensentation format, but i'm still a little bit confused about two (probably more, haha) things. I'm just learning about architectures, so excuse me, if my questions seem obvious:
- 1.) Let's start with the roboter vacuum cleaner. I'm having trouble understanding the flow of data and corresponding actions being taken. The roboter sends an event to the event backbone (a kind of server I assume), which triggers a FaaS, which takes action "with the vacuum cleaner" eg. cleaning the house? How is that achieved? Does the robot have to have a connection to be able to clean?
- 2.) Store case: A customer triggering a checkout event, which updates the inventory seems logical. But how about responses to user specific data. E.g. denying the checkout because of an empty inventory or a payment (verification) event, which only corresponds to one specific order. Same thing can be asked about the spotify play event, which triggers the "suggestions" process. Who handles sending the song data to the client or the suggestions, which only apply to a users currently hearing that song.
I would be glad if someone could sacrifice some time and offer their insight. Thanks!

cyk
Автор

Curious to know how you are able to do that mirror thing. 🤔

SyncWithSrikanth
Автор

Thanks for all of the great content lately.

ftate
Автор

"Statement of fact": almost 30 years ago, Sybase databases (dont know about others) kept writing something that was called "transaction log". That e.g. made it easy to implement replication. Transaction log is about logging all data manipulation. Statement of fact. Every single sql-statement executed.

jootuubanen
Автор

I feel like this class can only be understood by people who already know this stuff. Its a whole lot of names and acronyms without enough explanation of what they're meant for. Like event backbone for example. 7 mins in still there's no moment of clarity I was hoping for.

asishreddy
Автор

There's something about using terms in this video that I do not fully understand. In the introduction the presenter shows that message driven is one of the key points to achieve scalability and resiliency. But then, it shows that that's achieved with an event driven architecture and not a message driven one.

marcoberlot
Автор

It's great, in starting I found everything is bounced off, but later got a crystal clear understanding. Thanks for making it simple & clear. One question comes to mind what are the other architecture, can you name them...!!!

the_nobody_entity
Автор

The checkout example is one of the worst usecases chosen for event driven systems. An order has to be coordinated, consumers would need to sync because an order has to act like a distributed transaction/saga. If you would implement it like described in the video the customers will be contacted the shipping left warehouse, but the product was not even on stock! The order has to be done in sequential steps which can fail or get canceled, by having a queue system you will lose the ability to do so, and each consumer will react async in both time and space.

curiosull
Автор

Stream history is one of the biggest challenges in an EDA / microservice architecture. Not all services start their existence at the same time. Or even worse: You want to integrate legacy systems with events originating from them - but they didn't produce events from the beginning of their life. So it's much easier to talk about an EDA-based system if you just see the present. But how do you make sure every (new) consumer has seen all the data it needs - originating from events?

marcom.
Автор

Its quite easy to understand, please make more such videos

swapnilkulkarni
Автор

thank you very much, I like IBM company

rezabojnordi