Cloud Native Computing Foundation OTel Community Day NA 2022, 29 Jun 2022

Previous Meeting Next Meeting

⏯

youtube image

►

From YouTube: Lightning Talk: MARA: An Easily Deployable Modern Application Architecture - Jason Schmidt, NGINX

Description

SAVE THE DATES!
ArgoCon – Register Now | September 19-21, 2022 | Mountain View, California - https://events.linuxfoundation.org/argocon/

KubeCon + CloudNativeCon North America | October 24-28 | Detroit, MI - https://events.linuxfoundation.org/kubecon-cloudnativecon-north-america/

KubeCon + CloudNativeCon Europe 2023 | April 17-21 | Amsterdam, The Netherlands - https://events.linuxfoundation.org/kubecon-cloudnativecon-north-america/

OpenTelemetry is a collection of tools, APIs, and SDKs. Use it to instrument, generate, collect, and export telemetry data (metrics, logs, and traces) to help you analyze your software’s performance and behavior. Go here to learn more: https://opentelemetry.io/

Lightning Talk: MARA: An Easily Deployable Modern Application Architecture - Jason Schmidt, NGINX / F5
303+304 (JW Marriott Austin | 110 E 2nd St, Austin, TX 78701)

Speakers: Jason Schmidt
Link to join the virtual event: https://www.accelevents.com/e/otel-community-day-na-2022
The virtual event concludes at 1:50PM CDT; In-person event concludes at 5:00PM CDT

MARA, the Modern Applications Reference Architecture is complete Kubernetes-based application deployment that includes observability as a first class citizen. MARA uses Pulumi as an Infrastructure as Code provider and can provision Kubernetes infrastructure to a cloud provider, or use an existing Kubernetes deployment. The authors are using MARA to explore the process of adding OpenTelemetry to an existing application / environment. Currently, the OTEL operator is used to deploy a collector which is used for traces, with future plans to leverage OTEL for logging and metrics collection.

A

uh My name is jason schmidt. I am a solutions architect with nginx and I'm here today to talk to you about a project we've been working on for about the last year that we call mara and what is mara. This started with a deceptively simple ask: we were asked: what is a modern application? Could you help us define it and, as we started working on it, we realized that a modern application is in many ways like an iceberg.

A

We have the application at the top and that's what we focus on, but there's a lot underneath that is needed to make that application work. There's monitoring, observability management somewhere, there's a kubernetes installation somewhere, there's actual hardware, so mara was born and our one liner is: we've created an example architecture using kubernetes that aims to be as production ready as possible.

A

So who are we? And why did we do this? I'm part of the community and alliances team and we have two mandates. We work with the open source community and we work with our alliance partners and in a lot of cases there is overlap. So we work with lightstep and rancher and jfrog and they have a foot in both camps. They have a foot on the open source side and a foot on the alliance side.

A

Specifically to my team. We have a very diverse array of technical experience on my team, probably one of the most technically deep teams. I've worked with- and I put some of these experiences up there not to impress anybody but more to talk about the different viewpoints we bring to every problem that crosses our desk.

A

Anybody who's tried to get dove people and operations, people and qa people and managers all together on the same page know it can be like hurting cats at times. So uh this is something we're all happy with, and there have been lots of discussions and arguments to get to this point.

A

So why did we build this? With all the experience we have and given the mandate to explore what a modern app was? We wanted to leverage our experiences to build a best practice example, because we couldn't find one out there. We wanted a way to deploy kubernetes infrastructure that did all the things we needed. It to do to support a modern app and in looking for something to fit that bill, we found a lot of anti-patterns marking materials masquerading as technical artifacts toy applications. That really didn't do anything.

A

My personal favorite, the long demo that doesn't actually work when you try and put it on your own machine or your own system. I'm sure we've all hit that and then of importance to our team is to highlight open source as a core part of modern applications, not a bolt-on, not an add-on, but a core part.

A

So, what's in the box, what did we put in with mara as you look at this? I don't think, there's any surprises. Here we run on the major cloud providers we run anywhere. We can have a cube, config file ingested and get persistent volume, support and egress ip we've chosen to use pollumi as our infrastructure and code per infrastructures code provider.

A

We use the python language and we use plummy because we felt it was easy to read and something simple for us to use. Observability we're going to talk a little bit more about our application. We chose the bank of anthos, which we forked and called bank of sirius.

A

Our desire was always to add more applications to this and today's blog post from the hotel hotel foundation. We do have something to look forward to because we'll be working with that to make it work with mara.

A

So, who runs mara uh we've got some sample use cases and out there on our team, the folks that are developer heavy. They just want a way to stand up their application. They want to put their application out there and have all the plumbing taken care of.

A

I come from an operations background, I'm interested in how we manage, monitor, deploy and support these applications. I also like to break things and see how we recover. What's our mean time to recovery, what failure modes do we induce when we break components, qa engineers on the team they like having a reproducible test environment for benchmarking for testing and then the most interesting use case? For me, it's labeled as startups here, but this is anybody who has a component that they want to test out in a running application framework.

A

One of our first integrations outside of what we put in mara was with sumo logic. They came to us and said they wanted to plumb their solution in we turned some things off put theirs in and had it running within a day and to me that's the strength of mara. We can plug and play different components and see how they act within the mara framework, so we're at an open, telemetry conference. How does otel fit into mara now we're not far on this path.

A

Right now we have an hotel collector that consumes traces from the bank of sirius application and then we're using elk for logs and prometheus for metrics we visualize in grafana we visualize in cabana and the operators we use the hotel operator. We have a prometheus operator out there in elk, so this is now and we've.

A

Our first step was to get mar up and actually running now we're at the second step, we're starting to layer hotel in. Where do we want to be in the future?

A

We want otel to take care of metrics logs and traces.

A

We want to figure out the best way to do that and the way we can do it where people can drop in their own solution, light step, be that jaeger zipkin, whatever you want, drop it in there and there's some question marks here, because we really don't know what the future holds: we're kind of riding along and seeing what comes out and what works.

A

So that all said. If we've reached this point, you're thinking man, I want to try this out. It's very simple. We have a van d url nginx.com mara, clone. The repo have credentials for supported cloud, have credentials for a k8 installation that meets the minimum requirements. There's a setup script. You can run that will go ahead and set all the right versions of the software. We need there's a startup script. The startup script churns for anywhere from five to thirty minutes and at the end you're running mara.

A

Once you have it up, what can you do? You can log into the bank of sirius application?

A

You can generate a load with the locus load generator, which has some pre-written test plans that google came up with for bank of anthos. You can explore the deployment whatever tool you want use, cubecontrol k9s use the dashboard.

A

You can use the managing and monitoring tools that we've built in there's some convenience scripts. To help with that, you can change component settings, put your own components in and you can break things and see how it responds.

A

So, what's next we're looking at adding some better load and performance testing, we have a project in the works called cicada swarm that is going to. You know, hopefully add another layer of testing with some pass fail for ci pipelines, we're looking at migrating to the plummy automation api to get rid of the bash scripts that currently start things up and tear them down, and we're looking to have better configuration management of interest to this group.

A

Our goal is to track and adopt hotel enhancements as they come along, extend hotel to other products within the nginx organization and greater f5 organization and provide feedback on our experience through blog posts through conference talks, and things like that.

A

So, if we've gotten this far and you want to get involved, you can go clone the repo. You can raise issues you can participate in discussions. We appreciate honest feedback. Even if it's you're telling me this doesn't work, we have problems with it or you think of a better way. Everybody on this project is there's a lot of humility and we know we don't know best.

A

uh Prs are always accepted and we do have a community slack that you can join there. There is, or will soon be, a mara specific channel. You can drop into and go ahead and talk to us and then we're at conferences like this so stop by and see us I'm gone tomorrow, but I have a colleague dave here. Who's speaking later in the week, uh go bother, dave dave can get information to me or just join us on slack.

A

Above all, you know. This is something we did for the community. uh Nginx graciously allows us to spend the time to make this work for the community, so we would like this to be useful. We'd, like people to share the code and uh rising tide, lifts all boats, as we said earlier today, and with that I thank you for your time.