Welcome to Serverless 2.0 • Alex Ellis • GOTO 2019
Вставка
- Опубліковано 31 лип 2024
- This presentation was recorded at GOTO Chicago 2019. #GOTOcon #GOTOchgo
gotochgo.com
Alex Ellis - Creator of OpenFaaS, Engineer, Speaker and Author
ABSTRACT
This talk from the creator of OpenFaaS and CNCF Serverless workgroup member Alex Ellis explores what hype actually means for our industry and why it matters so much for serverless right now. You’ll learn why ThoughtWorks says multi-cloud portability is best achieved through containers and learn more about how the industry is shifting towards Kubernetes for container management. This has lead to what Alex is coining Serverless 2.0 - a truly portable experience built on battle-tested technology.
Alex will then explore the current landscape of Serverless with Containers and compare three of the most popular Serverless frameworks for Kubernetes. He’ll then explore two real-world use-cases before going on to give interactive demos where you’ll participate live with a Single Page App connected to managed Postgres.
OpenFaaS had its genesis in 2016 and has trended as the top open-source project on GitHub, won Best Cloud Computing Software from InfoWorld for two years running and has a thriving community with over 190 contributors, 3k commits and over 16.5k stars. A growing end-user community of businesses and teams are taking serverless into production with OpenFaaS and Kubernetes [...]
Download slides and read the full abstract here:
gotochgo.com/2019/sessions/728
/ gotochgo
/ gotoconference
/ goto-
gotocon.com
#serverless #OpenFaaS
Looking for a unique learning experience?
Attend the next GOTO Conference near you! Get your ticket at gotocon.com
SUBSCRIBE TO OUR CHANNEL - new videos posted almost daily.
ua-cam.com/users/GotoConf... - Наука та технологія
One key thing not mentioned about Serverless 2.0 is language agnostic support. This means using a Serverless platform that can automatically support any programming language without the burden of the language runtime requirement (all proprietary Serverless platforms have this requirement - AWS Lambda, Google Cloud Functions, Azure Functions etc). Language runtimes create unnecessary maintenance/performance overhead, and are tied into a specific language version which hinders using newer language versions, and can even prevent certain language features from being used.
What is great about OpenFaaS is the flexibility to use whatever infrastructure you already have or plan to deploy You package your functions or services into vendor agnostic containers that can be built even for multiple architecture like X86_64 and ARM. Adding to some of the comments, it doesn't require Docker. On the container building part, you can use whatever tool you want like Docker itself, Buildah, Makisu or something else as long as it's OCI compliant. On the server side, OpenFaaS can be deployed into Docker Swarm, Kubernetes and it's enterprise versions like Openshift or even managed clouds like EKS from Amazon or GKE from google with one single command. In the end what matters is focusing on your application.
and at the end the client will ask u, are u done yet?. and the domain will be left fragile, ugly, stuck on infrastructure level.
Folks may also enjoy the talk that this follows on from, more introductory content in Serverless Beyond the Hype: ua-cam.com/video/yOpYYYRuDQ0/v-deo.html
This man is going from extreme high level serverless concepts directly into an openFAAS cli without any explanation. Using docker is a mayor extra pain in the bottom and the presenter doesn't touch on this at all What is openFAAS offering that makes it a better experience compared to say using deployment tools, like the serverless framework, with AWS?
Sorry to hear that you feel that way Sander. Why don't you look up more information about the 2.0 concept on the blog or docs? www.openfaas.com/blog/
Let me know if I can be of any help to you.
What troubles me is that running and managing docker images in a Kubernetes cluster in production is neither easy nor is it free. "just add Kubernetes" is not a solution to get rid of your operations, it in fact adds a lot of operations work to your plate. This work is not needed when you go all in with a vendor, be it AWS or Azure. Now I understand that some use cases need this vendor agnostic approach, but in my opinion most of them do not. If I see your talk you seem to have the idea that going with OpenFaaS relieves you of a lot of work. I now have experience of running a full serverless stack in production on AWS for over a year and I am not running into any of the problems you seem to think are there. I actually enjoy it a lot and would recommend it to anyone. Again what do you feel is making life easier by adding more operations work needed to run OpenFaaS?
@@SanderE39 If you are an AWS customer, and completely happy with all aspects and limitations of the SaaS product, then you're the first type of Serverless user I mention. You should absolutely stay on Lambda. It doesn't sound like Kubernetes is appropriate for you. You may find that many other companies are running Docker and Kubernetes in production. You can find out more from the CNCF. www.cncf.io/projects/case-studies/
The "offer" that OpenFaaS provides is removing the limitations of the Serverless 1.0. CPU and memory limits can be adjusted to allow for long running jobs, or training machine learning models. The development process is familiar to both developers and operations in that Docker containers are the unit of operation. Also, the choice of language, and version of that language, is near limitless, enabling more organizations and teams.
Some have fully bought into the Serverless 1.0 model, and adjusted their development process to work within those limitations. Others find that the limited features, and radically different operational process hinder their ability to utilize the benefits of a serverless environment. This is where Serverless 2.0 comes in to play. As mentioned in the talk, it's not just OpenFaaS attempting to help relieve these limitations and provide a better experience for both developers and operations teams.
@@burtonr When I read your comment I get the feeling you have never actually worked with AWS and think serverless is just about the functions or Lambda. What makes AWS so enticing is the fact they offer a whole skew of services to handle, for example, creating, training and working with machine learning models. This is not limited at all by Lambda inability to run for maximum of 15 minutes. These are things people fear when they don't have a good grasp on what serverless actually is. And I think it's not about running one function for hours on end. Serverless functions should be small dedicated pieces of code whom you can string together for longer workflows. Something AWS also has some great tools for, like step functions. All this and much more can be used without ever configuring or maintaining one docker image let alone a kubernetes cluster. Besides this our hosting cost are laughably low, especially if you include the fact I don't have to hire expensive Kubernetes experts to rangle my production environment. So I can spend my time creating more value for my customers. I don't see any real practical reason for going this serverless 2.0 route just because: more memory and longer runtimes, really that's it? You are just repeating your point, if it is such a surefire way to succes why can't you use arguments to persuade me. I feel like I am missing something here.
In principle, I want to love OpenFaaS, but it just doesn't feel like it would improve my workflow. The Serverless framework has already eliminated a lot of pain points with AWS Lambda, and it wasn't clear how OpenFaaS would make my development workflow legitimately easier. As an example, trying to get a function to run with a specific IAM role? With Serverless framework, you just specify the ARN in config or let the framework deal with it. In OpenFaaS, the whole thing is built around agnostic containers running in k8s. Great, okay, so now we have to get EKS fired up and correctly assign IAM roles to each pod. That's fine. But OpenFaaS isn't going to help you out, because the entire concept of IAM is an AWS-only concept, so it kind of leaves you in the cold to figure it out on your own. Need to actually use AWS lambda under the hood? Bummer, because faas-lambda is in "early preview" and isn't actually documented or ready for prime time.
Sometimes we spend a lot of time being afraid of vendor lock in, and not enough time thinking about the day-to-day problems of actually getting code to work in the trenches.
In my talk I mention that if people are happy with being locked into a single vendor, then they should stay on a Serverless 1.0 solution. It sounds like that's you, but it's not everyone. An example might be LivePerson for instance -> ua-cam.com/video/bt06Z28uzPA/v-deo.html
Serverless 1.0 vs 2.0 is not only about leaving lock-in behind, it's about being able to loose oneself from all of those limitations and being able to take advantage of everything that Cloud Native has to offer. If you'd like to know more about faas-lambda, subscribe to the blog where you'll find out more -> www.openfaas.com/blog/
Let me know if I can be of any more help to you, Alex