I got a Target Dojo sticker from him at DOES in 2015 and had it on my Walmart Laptop until I had to use my laptop to present a DOES in 2017. 😄

Hope to connect in the Dojo Consortium. 😄

I've been very inspired by the dojo work presented and we are shifting out Engineering BootCamp into more hands on learning

We're looking for ways to scale hands on learning. This is a really good demo.

We split DevOps Dojo in 4 pieces: • The white belt DevOps Dojo is the "what": understand what DevOps is and how it is different. • The Green Belt is the "why": for leaders (we all are leaders according to Jonathan Smart!) and executives, so that they can understand the dynamics and support DevOps transformations. • The Yellow Belt is the "how": it goes through various patterns, mostly articulated around the Accelerate book.

The Online DevOps Dojo is our Open Source version of the Yellow Belt DevOps Dojo. Finally, we have a Black Belt DevOps Dojo, which is what is the closest to the other Dojos I know of (from Target, Verizon, Delta Airlines, Walmart, ...).

Do you have a link?

We are trying to implement something similar.

https://dxc-technology.github.io/about-devops-dojo/

Danke

@gvian, right. Feel free to actually go through it in 4K: https://dxc-technology.github.io/about-devops-dojo/modules/

Not helped by the windowing around our video (which was recorded at 1080p)😢

one more reason to explore after the conference has cool down 😉

GitHub and GitLab are both looking to address this leveraging Visual Studio Code in the browser

it's the Codespaces beta, isn't it?

I've not seen that.

it's a new GitHub offering ping ppl on #xpo-github

#xpo-github - you may be interested in connecting

MS Learn has many of the same features as Katacoda, but also great integration with Azure, and I'd also expect great integration with VS Code Similarly Google have Qwiklabs, which is integrated to their stuff.

Oh yes, I remember that now.

Thx!

@matthew.cobby, you can have a look at https://github.com/cdr/code-server

Thanks! Our internal prototype took this code and deployed in our cloud context. Security had challenges with it so put it down for now but will go back to it

Killjoys

We've had >26,000 people through the entry level 'white belt'

and yes, there was a push from the top to do that. When we first started one of the things I asked the EVP of delivery to do was to run through the training himself so that he could show leadership from the front

Impressive! How do you keep record of belts? Badging platforms? Excel 😂

he came back to me with some wrinkles on using Katacoda on his iPad 😕

How are you tracking usage?

@matthew.cobby we host the training in our learning management platform (Sabacloud), and @olivier.jacques drove the creation of badging capability

These modules are in our LMS / Learning Management System, this is how we track. Now, we (DXC labs) also created a very lightweight badging platform (serverless), which we may leverage to gamify all this even in the open source version.

@bryan.finster we get reporting out of the Sabacloud learning management system

How many belt levels?

@pgibbs1587 white, yellow, green, black; and we broke yellow (that this module comes from) into multiple stripes

@pgibbs1587 - https://devopsenterprise.slack.com/archives/C0150HQB6UX/p1592994609186000?thread_ts=1592994568.185800&cid=C0150HQB6UX

We've a similar idea, but hadn't solidified what they would represent. Do you also badge teams for outcomes?

We're looking at using delivery metrics to do that.

@bryan.finster we haven't come up with a team badging approach, but that's an idea we should probably dig into

I think both are very valuable.

Whats the time frame it takes to go from white to black belt?

Its the gamification of badging we are looking to tap into. We can integrate to our LMS but it's an Oracle system that is hard to use and no-one goes to unless they have to We'd like to build a tool like the AWS Phone Book

Something that people see and can help find SMEs across the org

@pgibbs1587 at least 6 months to get to black belt, but usually longer than that. At one stage we were frequently running on site 'buildathons' for one of our big initiatives, and that allowed us to grow black belts a little quicker

@matthew.cobby yes, we have an internal "team roster" that's used to glue many things together with HR data. Something like that helps.

@matthew.cobby for some other stuff we were also using Open Badge Academy before it shut down, which we've now had to replace with an internally developed DXC Badger

Are you using the katacode usage metrics for your badges?

@tommy we do look at the Katacoda usage metrics, but they're not used for the badges

yes, i think we can commit to that happening.

None. All the modules are provisioned dynamically. It is Docker based.

Jenkins runs on top of the Katacoda platform, so our only external dependency is GitHub. We use our enterprise GitHub for the internal Dojo with the full range of modules and public GitHub for the open sourced modules. It could be easily adapted to use another enterprise GitHub (or GitLab, Bitbucket etc.)

I didn't realize that. Brilliant.

@olivier.jacques @chrisswan Love to catch up after all this is over for a chat and compare notes

Katacoda is all based on containers, so we can glue multiple containers together to create a full environment

@matthew.cobby we can find a time to connect - I'll DM you my work email

Yep — outcomes are things like popularity & security. We give more detail later in the talk.

Haha 😂 this is super neat.

Yep, that's pretty awesome, getting to hang out with Rich. Dude's a legend! ;)

Sensei Rich

Good seeing you, buddy!

Loving this one Gene ... I like how you are moving at a digestible pace. I remember vividly having a conversation with someone about "Maven" where the person across from me was completely lost for 5-10 minutes. Then he said "Oh! Like npm! ... Now can you start over?"

I was looking about the framed certificates at your back and thinking about the size of mine, poster-size

Yeah, I think CMU tried to win the “largest diploma” award.

When my family was moving a few years ago, the movers told me that they could move my wife's U of Miami diploma but not my Rowan University diploma (too small) 🙂

Awesome. I've been pretty non-technical the last 2 years, but have been diving back into Clojure on the side, as you challenged "not staying technical". Been hard, but a blast! Loving Re-fame.

YES!! Love re-frame — but wow, it took a lot of reading before I actually knew what to do! 🙂 So delighted that you jumped into Clojure — it really changed my life.

Oh yeah, I had been learning Clojure a few years back. But yes... still plenty of reading and watching Eric Normand, before any progress. Lol.

Eric Normand FTW! 🙂

I really want a clojure that runs on the BEAM VM

Total WTF!

Yeah, it was surprising. I will say that dataset was sparse. There were not a lot of projects in our dataset with github repos with Travis or Circle CI configs.

I would say it indicates either 1) the effect is not huge or 2) there is some substantial use of CI that we weren’t capturing (e.g. separately-managed infrastructure that is not provided by Travis or Circle)

Software is an asset .. code is a liability!

I’ve come to peace on this — to me, it’s a measure of batch size. One of my favorite metrics from SODR was commits/dev/day, which increases linearly for high performers, flat for medium performers, and goes DOWN for low performers.

Does it apply to all types of work? For instance for ML or AI work you might be a high performer even though your commits are low

Steve sees the arrows before they happen on screen. Anticipation!

So... you feel a need...?

For ... speed

🙌:skin-tone-2:

LOL

:high-five:

depends on what tooling you’re using to check security of dependencies

it might be just marked as a vulnerability or maybe a “transitive vulnerability” (or similar term)

but update performance of libraries is not obvious at all just by looking at the repo / code / etc. it needs to be computed from historical update data which is not easy. still working on a good way to fix this and provide all that data to everyone. for now we have a list of the top projects in terms of update behavior in the report.

Have a direct link @stephen?

@dacahill7 jus to echo @stephen this is a challenge most organisations face as its a real struggle for all Developers and Maintainers. Ideally PPT (People, Process and Tools) still applies and its understanding what that vulnerability is so that you can either remediate, upgrade the direct dependency or simply use something else. This is across Dev and Security, but ultimately you want to make sure you have the right tool to identify if it is a True Vuln or not. Take a look at some of the things https://www.sonatype.com work on to get an idea of the best way to tackle this https://www.sonatype.com/wp-developer-demo-book

Yes, there’s a substantial middle ground between the exemplars and the laggards.

@richard431 said you have a cloning machine!

Yeah, this is something you can't do in-person... I wonder if continuing an online virtual conference remains in the cards, "post-covid".

I wish our elected officials would do this

Was just talking about that with @mik last night. I suspect our physical conferences will definitely be different in future, given this experiene!!!

Yeah.. Makes a lot of sense. #Datomic

Looking forward to seeing if you use Datomic on the next one.

Have you got to use it any further?

it's all in the community of people behind and I do not think that Javascript community shines in good engineering practices

a focus on short-term solution ex. I was shocked by React idea of building/packaging for a specific target environment it implies that what you test in environment X is different from what you deploy on env Y

Hi Daniel, yes - let me set up a quick call with one of my colleagues to discuss this. 🙂

👍

I wonder how we could gate on that?

IMHO there are many factors to consider one is where a project lies in the cycle e.g. older projects might have lots of popularity but little changes and core maintainers moved to new interests

another is the tools that help/support in managing dependencies how automated is the flow from CVE to updated dependencies and how much is manual?

We recommend avoiding complicated subsystem teams when possible. If you can provide those functionalities in a self-service type of platform that's much better.

Makes lots of sense. Did they manage to get CRM as a self-service platform? That’s an awesome aspiration.

Otherwise you can easily get into component team anti-patterns / dependencies / Bottlenecks

Of course

Oh interesting here that CRM is a stream-aligned team

I found that teams using CRM chose a bad packaging at the start, splitting along technology instead of business features

So a Complicated Subsystem is sort of a mini-platform with a single purpose/focus. Certainly, many of the behaviours around each are very similar.

Yes indeed - I’ve seen CRM done badly many times, and well very few. The technologies don’t help - Dynamics, Salesforce, etc.

they have a security layer, a data layer, a UI layer shared by all teams instead of vertical slices (yes, some cross-concerns are ok but very limited)

I saw this multiple times with SharePoint, Dyanmics it's not the platform it's the community that have lots of business and data knowledge but little Dev & DevOps background

anecdotal but when you see a pattern repeating across companies and technologies... you know it nails something

100%, and - for example - configuration management in Salesforce is almost non-existent.

Yes as we began to break the monolith into smaller parts and therefore separate into repositories where appropriate

Workflow management maybe a platform?

e.g. JBPM or one of the expensive commercial ones?

Avoid "component teams" 🙂

UI components we have as an enabling team

(And maybe even trending towards a platform)

i have many teams that end up building their own workflow management, their own UI components, etc… is that just an innersourcing problem?

That might happen because they don't see their needs addressed by other teams.

Platform and enabling teams need to keep at top of mind that their goal is to reduce cogntive load / effort of stream teams around aspects not directly related to their business focus

yeah… i think it has come about now as the default position and we don’t explore what other teams have done… probably a different problem than team structure 🙂

Often the problem is about how teams interact and understand their purpose

I’ve been preaching inner sourcing within my area of focus… if another team’s components do not meet their needs, bias towards inner sourcing and contributing as opposed to building your own.

I worked with John Kilmister and was trying to get them to move to product/stream aligned teams for a while, but when the TT book came out it cemented some of the concepts and provided the catalyst to make the change

BEst ask @richard.allen who led many of the activities 🙂

🎉

So it sound like you had an idea of what would be helpful already @richard.allen and then the book helped by giving a framework and a common language… is that right?

We rather avoid calling it a "framework" :)

🙂

er… a model!

@jtf yes that sounds about right

The way Rich Allen told it to me was that the book arrived at JUST THE RIGHT TIME 🎉 - an on-time delivery 📦

+1 to that

Nice! 🙂

Depending on the business goals each different team might be impacted in different ways so there may be times where the teams to need to collaborate for a period of time

My understanding is that part of the reason for taking 3 months to look at the streams was to find good streams and so avoid the need to switch priority often between streams. Is that right, @richard.allen?

@richard.allen so teams also have to work out of their "own" streams, having to know other areas also?

Yes, certainly the intention is to avoid the constant switch in business focus that seemed apparent when using project and BAU teams

@tfr the teams "own" their streams and the products within their stream but they would only need to know how to use the X-As-A-Service published by the other platforms etc - goal is to decouple the teams as much as possible

@richard.allen makes sense, I have the same situation with my 7 DevOps teams, they have own area, but has to be able to work in "neighboring" areas, main focus is to keep stable team. The tricky part is finding those good streams that are long lasting with constant flow of work from business

To find good streams for fast flow of change can be hard - needs new skills and ways of thinking. Some places to start: 1. Independent Service Heuristics: https://github.com/TeamTopologies/Independent-Service-Heuristics 2. DDD Starter Modeling Process: https://github.com/ddd-crew/ddd-starter-modelling-process/ 3. DDD in general: https://dddcommunity.org/learning-ddd/what_is_ddd/ / @tfr

Thanks @matthew, just bought your book on my Kindle, looking forward to reading about your findings 🙂

Also we have a public training session tomorrow and friday that covers this: https://teamtopologies.com/events/2020/06/25/online-training-team-boundaries-and-architecture-for-fast-flow-tt07-25-june-2020 https://teamtopologies.com/events/2020/06/26/online-training-team-boundaries-and-architecture-for-fast-flow-tt07-26-june-2020

Yes the teams tend to be around that size

That's why we included it 😂

Too many good quotes in this preso... this is golden as well

QDPD quote-driven presentation development 😂

Yep, absolutely. The Enabling team type is there to do exactly that.

In Visma we have a team of 10 people who spend all of our time on helping teams adopt and improve practicing modern engineering practices (Continuous Delivery, DevOps, Public Cloud)

Nationwide does this too - we have a Ways of Working team

Definitely. We expect that as teams become more independent, the traditional middle management will take a boundary-spanning role, helping teams understand expectations and improve interactions with other teams.

@jonathansmart1 had a similar role at Barclays, if I'm not mistaken

This is the job of leadership surely?! An agile coaching practice can help them (enterprise coaching, team-level coaching), but not a CoE type of structure.

The enabling team in your book is more a temporary team. Or did I got this wrong?

Agile is not an expertise...

@akis.sklavounakis - that's a good point. We actually recommend that some managers change their focus and - instead of managing people's time and tasks - they focus on setting up the right conditions for other teams to learn and share practices effectively

The Enabling team has a temporary relationship with another team, that's certain. An Enabling team may be short-lived or may be long-lived. They thing is to avoid creating a dependency on the Enabling team

@matthew So not checking up on people working remotely I take it 😄

Sure hope not @akis.sklavounakis! It's not about how an individual team works, we expect them to have sufficient autonomy + accountability.

It's about helping different teams better understand how to interact, when and for what purpose.

@rene.lippert we're discovering interesting variations on Enabling teams.

One key aspect is the domain of expertise of this team, e.g. Continuous Delivery is a wide domain, expect an enabling team to be long lived inthat case.

Vs something like test automation for mobile, which might be shorter-lived

@me1208 I am still unsure if I should for example put all my disciplinary mangers into an enableing team.

I can't answer that without more context 🙂

We understand them as coaches but they also make all the decissions where money is involved eg. training or hiring or ... what you have in the old company org style thing with team leader, bussines unit leader, ...

That sort of responsibility is quite far removed from the behavior and intent of Enabling teams, @rene.lippert I would be very careful about mixing budget responsibility and decision making from Enabling activities ⚡ ⚠️

That said, we typically see the need to change those aspects as well, from an individual-focus to a team-focus (training is an obvious example, but even hiring needs to start with team and individual purpose alignment - see this article: https://techbeacon.com/devops/why-you-should-hire-devops-enablers-not-experts)

Thanks! Very interesting article. I think I still have to break it further down. Stay in touch.

No, they work with the stream-aligned teams to help them reduce toil as they are responsible for runtime of their service.

There is no such thing as "SRE infrastructure" really 🙂 There is infrastructure that the SRE team may help to define and diagnose, but that infrastructure is ultimately owned by the Stream-aligned teams

But they also act as a "sensor" to common problems across stream teams that the infra platform could possibly help with.

They are like a "bridge" between stream-aligned teams and platform teams in this case.

But their role is enabling, teaching what SRE means and how to handle service reliability, performance, etc more effectively by the stream teams.

Understood. Is then the infrastructure for e.g. detecting SLO breaches used on a fully SaaS basis by all the teams?

The cloud platform team provides a service that, based on a couple configuration files provided by the stream team with SLI/SLOs, creates SLO dashboards per service

It's also integrated with their chat tool, so the stream team gets notified when certain thresholds are crossed.

I see. Thanks for the clarification. I thought that someone in the org has to own this 🙂

Good point.

It's one of those cross-team concerns that any individual team will find hard to justify to invest on by themselves.

SRE team helps surface the need, then platform team developed the needed service.

Platform includes other such services like dynamic databse credentials, and other security related srevices

Right. Like e.g. Secret Rotation Service?

Could be, Vlad - there would be very little value in most Stream-aligned teams from managing their own secrets rotation 🔑

It depends :)

If you can outsource your platform to cloud providers that’s awesome

As a regulated company we can’t quite do that

So we have to add a thin layer of platform - what Team Topologies call a Minimum Viable Platform - on top of our cloud provider

But if you can get away with having someone else do your platforms for you then 🥂

we're kind of trying the same, building our services on top of Azure PaaS as much as possible. Allowing the teams to use almost whatever services they need from Azure

🙂 Hi! (from the Uswitch one)… we did it because we can have a small number of people doing high leverage work. Some of that is work that will ultimately go the way of our cloud provider (AWS) which is awesome, but most of their value-add is higher-level/value than what we’d expect AWS to add to a roadmap

This is really important thinking and practice

https://medium.com/@pingles/convergence-to-kubernetes-137ffa7ea2bc had some more detail on rationale/impact of that work

And I should say stream-aligned teams are still responsible for lots (creating DBs etc.), but they can do so with higher-level automation/abstractions than they would otherwise

but that also comes with a cost - there's quite much cognitive load for the teams to sort out all kinds of Opsy things like backups, monitoring etc. Now that teams use a lot of various Azure services, its very difficult if not possible to have a platform team that would take over responsibilities from the teams

It seems to me that sometimes you want to use development competence when performing operations, and operational competence when performing development. Hence responsibility for both in one team. A centralized platform team can not have great knowledge on the 50+ applications (and the customers, domains, etc) running on their platform. We do have teams of advisors that can help our stream aligned teams follow good cloud architecture practices, this helps take off some of the load, but far from all of it of course.

There should be an ongoing assessment of the cognitive load on Stream aligned teams. As technology from cloud providers improves and increases, organizations should be looking to adopt the new services into their internal platforms and throw away / shrink their previously in-house versions.

How many of those advisor teams do you have? And how big each of them is? @alexander.lystad.

At NBS, in Digital, we built a platform to cut the cognitive load on the team -- so the stream aligned teams can ruthlessly copy what works from their peers. But the teams can also innovate should they choose. The great thing about using the 'approved' platform is that it's been blessed by the Security groups so you're in safe territory if you use it....

@vladyslav.ukis a dozen advisors that covers continuous delivery, devops, public cloud, architecture, application security for 60+ teams (and growing)

our stream aligned teams vary from 2 to 50 unfortunately, but the average and median is around 8-9

"blessed by security" sounds dangerous. how do you make sure it doesn't become a crutch?

It's a simplification: we know that using the central platform short-cuts some security approvals elsewhere. (We're heavily regulated.) If a team knows they need something else that's genuinely unique then they can do that but there's then the additional overhead of involving the security folks.

i see

"Invite, not enforce" is a strong part of the culture

Great questions Karl!

All the infra at uswitch is on AWS, including everything in the platform.

Therefore they were able to use traffic direct from customer services to AWS vs traffic to AWS via platform services as an indictor of platform adoption.

Platform is using Kubernetes on AWS and thus providing higher level abstractions to the stream teams (Deployments, Services, etc)

Gotcha, cut out the middle-man so to speak, a better product consumption model, or maybe 'repurpose/refocus the middle-man' to provide better services, faster 🙂

Def! They realized how stream teams had "too much" autonomy meaning spending a lot of time on low level AWS services management themselves.

On your last point, def true!

Platforms are actually fractal, you would expect to have multiple stream-aligned teams inside the platform as it grows (e.g. "monitoring" stream-aligned team, "provisioning" strean-aligned team, etc)

Many thanks for your replies, and great talk, nice job!

We also start to see "higher level" platform services like data APIs that themselves use "lower level" platforms services like provisioning and monitoring.

Like a russian doll 😂

We talk about that in the book a bit and in our platform training: https://teamtopologies.com/events/2020/07/02/online-training-principles-and-practices-for-modern-platforms-tt08-02-july-2020

It's a great format, thanks to @genek101 and the entire IT Rev staff!

Thanks!

The Art of Business Value was the first - what does "business value" really mean? One of the chapters was about the CIO's role so I decided to write anotherr book on that. A Seat at the Table is my book for CIOs and IT leaders. I wrote War and Peace and IT to tell a similar story to non-IT executives. The (Delicate) Art of Bureaucracy drills further into bureaucracy, a topic which came up in all of my previous books.

Yes agree. One of the big problems of RPA (beyond the integration at the UI layer anti-pattern, brittleness etc etc) is that it appears to destroy the business case for fixing the actual problem - ie removing the process altogether/leaning it out/redesigning from scratch/proper systems integration/etc.

I say "appears" above because I see the value from sorting out the problem properly is usually under-estimated as is the cost of maintenance of the RPA solution.

they are just code like all code, they require love and care

https://sdd-manifesto.org/