Cloud Native Computing Foundation KubeCon + CloudNativeCon North America 2019 (San Diego), 22 Nov 2019

Previous Meeting Next Meeting

⏯

youtube image

►

From YouTube: Sponsored Keynote: Network, Please Evolve – Chapter 2 - Vijoy Pandey, Vice President/CTO Cloud

Description

Join us for Kubernetes Forums Seoul, Sydney, Bengaluru and Delhi - learn more at kubecon.io

Don't miss KubeCon + CloudNativeCon 2020 events in Amsterdam March 30 - April 2, Shanghai July 28-30 and Boston November 17-20! Learn more at kubecon.io. The conference features presentations from developers and end users of Kubernetes, Prometheus, Envoy, and all of the other CNCF-hosted projects

Sponsored Keynote: Network, Please Evolve – Chapter 2 - Vijoy Pandey, Vice President/CTO Cloud, Cisco

Connectivity, security, and observability are critical to developer productivity and application velocity. The Network Service Mesh (or NSM) Project attempts to simplify how multi-cluster networking is consumed in Kubernetes by extending the core concepts that are already familiar to the K8s community. This talk will demonstrate how easy it is becoming for developers to consume networking via NSM, when we change our 35-year old view of IP networking.

https://sched.co/UdLh

A

Good morning, San Diego, even though it feels like Seattle outside so back in Barcelona I, talked about the need for networking to evolve for a cloud native world today, I'm going to talk about how that evolution is critical, and it's already happening by looking at a real-world use case. But before I do that I have a question for the audience. Are you a real programmer now before you raise your hand?

A

Let me point out that the comic strip that you see behind me is a mash-up of two comic strips from xkcd and one from user friendly, and in that you see a programmer sitting at their desk and a person, approaches them and says nano real programmers use Emacs, and this begins a parade of people who say no real programmers use vim or cat or Ed.

A

But the clincher is when somebody walks up to them and says you guys are all wimps. Real programmers use magnets to edit the eye nodes directly on disk.

A

That is one viewport another viewpoint. It comes from Leslie Lamport who got his Turing award in the field of distributed computing. So he knows a thing or two about this word computing and he says coding should be the easiest part of programming and, if you're, having trouble coding, you're doing something wrong now. What does this have to do with networking, for one networks are a key component of any distributed system.

A

So let's take a look at the real world scenario and see how networks are programmed today in a cloud native environment.

A

In this picture up here we have a use case where there are four databases: three running on public clouds, one on Prem and they simply want to communicate with each other securely to be able to either shard and/or really replicate.

A

How would you solve this today in a cloud native world by the way, congratulations to Vitesse on graduating and becoming a project? So one way to program that network is to use magnets. So you start with the technical complexity of dealing with ever changing part IPs of dealing with different routes with different clouds, firewall VPN complexity, different configs and compliance needs for the different cloud, vendors and that's just technical complexity.

A

Then you run into organizational complexity where you have devs and develops and met offs and SEC ops and Colo providers and cloud providers, and you need to talk to all of them and each one of those labels is not a single person. It's a team, so meetings and tickets and meetings and tickets magnets.

A

Finally there's process complexity. Is there a process? Is it stable? Who is accountable? More meetings, more tickets, a developer's best friend not, and all of this is happening because the application architecture is changing in a cloud native environment and therefore the organizational architecture and the operational architectures are changing, and so you cannot use the concepts for physical and virtual networking, which will aggregate statistical networking concepts like switches, routers, v, switches, v, routers bgp to solve application level problems. We need to simplify connectivity paradigms application development, and this is where entered network service mesh.

A

So we are a sandbox project in CNC F we got a sandbox status back in April around the time of cucumber Salona, roughly 40 you contributors and Counting, and extreme interest from financial center prizes and SPS. As you can imagine, nobody wants to deal with magnets and we had a first ever day. 0 event, it was one hundred and eighty percent over capacity. So as a community, we are extremely excited about that.

A

So how do you solve that problem? That I just described using NSF?

A

Very simply, all you do is extend your CR D with a line that says: I want to belong to the DB, replication and SM domain, and then SM logically takes care of securely connecting all of those database parts and ensuring that only those database parts talk to each other.

A

Now, behind the covers, things are a little bit more complicated, but still pretty simple, where you have a network service endpoint running in every cluster, that takes care of the data plane between these database parts wherever they happen to be, and then there is an N SM registry that figures out the discovery process of the remote endpoint again wherever they happen to be, and all of this is done for the developer without them, knowing anything about ip's routes, subnets or any special CRI.

A

Now, if you want to look at a demo of this, this exact same demo is running the cisco booth. So please come by take a look there: demos on application, security and observability. You can talk to the NSM community reach out to us please, and then, if you visit our booth, you will get stickers with say magnets or NSF. Take your pick. Thank you.