Velocity 2015 – Highlights from Last Day

Entering the last day of Velocity. We – Andreas Grabner (@grabnerandi) and Harald Zeitlhofer (@hzeitlhofer) – continue to blog our highlights of those sessions we get to see. Here is the summary of Tutorial Day 1 and Day 2.

4:10PM: Microservices, Micro Deployments and DevOps

Alois Reitbauer (@AloisReitbauer) and Martin Gutenbrunner (@MartinGoodwell) from Dynatrace raising these questions to the people in the room:

  • “Who is scared when 5 teams are independently planning software deployments?”
  • “Who can deliver a complete overview of your running system within 2 hours?”

Interesting overview of what types of dependencies you should look at in your system especially if you break up monolith apps into many smaller Microservices: Gatekeeper (likely not very micro, complex app logic, very deployment critical), Single Point of Failure (Central App Services, Fragile to Scaling), Hub (“Legacy” migrations”).

They are proposing a “Monitoring first approach to adopting Micro Service, highly agile architectures”

They invite everyone to Work with them and Try Dynatrace and submit your feedback

1:45PM: Making CD reality at a large enterprise

Adam Auerbach (@Bugman31) from Capital One walking us through their last 3 years of transformation journey from Waterfall to Continuous Delivery. The initial stage meant 3 dev sprints, followed by merging all together, testing and then – well – fixing the things.

Establishing a System Team which provides all necessary services for Agile Teams (CI, CD, Automated Testing, ,…). They started out with CI. Treated that team as an Agile Team in order to provide these services and help other teams to use their services in their own teams.

He made it very clear that the adoption of TDD and ATDD was key for their success. I really liked his comment on “Manual Testers are now doing Automation. If they don’t have the skills yet they need to acquire them”. He hit on the importance of automated testing many times and gave some further input on how they do it in their rather complex technology stack.

Showed off his “Goddess of DevOps Dashboard” 🙂 -> nice to hear that they are going to Open Source this project.

Gave a good summary overview of the key pillars for successfuly Agile/DevOps Adoption: Collaboration, Automation, Fast Feedback (Stop the Line), Simulation and Virtualization, Controlled Rollout

Key Metrics: Output (=Velocity – will be replaced with Cycle Time), Speed (=# Change Orders), Quality (= # of Prod Issues)

Value Stream Analysis great tool to communicate ROI for leadership!

Final lessons learned: Reuse proven frameworks; evangelize at every level; be ready for the need of education: leverage dedicated resources; be prepared for additional costs

11:50am: Picking the right metrics

by John Feminella (@jxxf)
There is no reason to measure everything, so decide wisely what you want to measure. Properties like happiness and empathy are worth managing and therefore worth measuring. Humans are consuming the metrics. Therefore proper visualisation is essential.
Measures are also used for alerting. But a threshold is again defined by humans. And thresholds don’t adapt. They detect abnormalities, but not problems. A regular situation might look almost similar to a problem pattern. Therefore they don’t work.
What is a good metric?
  • simple: the metric should describe a quantity that is readily understandable by the people who will consume it
  • actionable: a change in the metric should be able to inform your response
  • credible: the metric should convincingly represent the property it claims to measure
  • reproducible : it should be possible to reproduce the metric yourself from the raw data, and the raw data should be obviously correct.
metrics are only one side of the coin, it’s essential to visualise them properly.
But there are some problems that come along with visualisation:
  • it’s easy to fool the visual cortex
  • it’s easy to manipulate data to tell the story you want to tell (to tell an arbitrary story)
  • can we trust our vision? It’s hard to show honest data, but it’s easy to lie about it.
  • it’s easier to compare lengths rather than counts.
  • make sure the right things are compared and comparison is easy.
John presented a couple of examples where visualisation went completely wrong.
Summary: interesting session, providing good ideas on what to consider when you start collecting metrics.

9:30AM: Keynote Session Highlights

Bruce Lawson (@brucel) from Opera Software kicks it off and amazing us with some incredible statistics to answer the question: “Where will your next customers come from?”: Its going to be people from Asia & Africa which are already doing much more business than we know! Backed by stats such as: Uber (1Mio Rides / Day) vs Didi (5 Mio Rides / Day) or Online Revenue on 11/11 in China was 3 times more than Black Friday & Cyber Monday combined.  This is your next target market!!

Many of these markets – the next billion mobile online users – are operating on lower end devices but want to access the same services: MAKE SURE your apps are optimized for these lower-end devices. Some new Web Standards that will help: Installable Web Apps, Device Optimized Images

Interesting concept of Proxy Browsers that Opera and others provide. These proxies render heavy pages and sends compressed images to the actual browser. Saves a lot of processing on the devices, e.g: battery and bandwidth.

Bruce’s Law summarizes this great talk: It doesn’t matter how smart your phone is, if your network is dumb! Check out his slides!

Patrick Lightbody (@plightbo) from New Relic enlightening us about his love for Burritos from Don Pedro but even more so his Burrito Button on his new iWatch which changed his live completely 🙂

Lesson learned from his experience greatly summarized in: Every business is becoming a software business

Summary statement: Fast isn’t enough anymore – its about delivering a delightful experience for the end user to boost business.

Ariya Hidayat (@AriyaHidayat) from Shape Security is en”lightening” us on how fiber optics allow us to enjoy surfing the web or video streaming without issues across the globe. I am not going to repeat what he said as this is clearly out of my league. If you are interested check out his slide material. Really interesting – and – he had a very good way to present this material so that it even made sense to me 🙂

Your Action Item: If you know a scientist please give them a hug 🙂

Mikey Dickerson working for US Digital Service became “famous” for his involvement and stories around Today he is sharing how he is “Interacting and Dealing with bureaucracy” and how it relates to everyone of us. Why? Because once we start interacting with a larger group of people – you are facing these problems that we refer to as bureaucracy 🙂 You got to watch him on the video – some interesting insights on

Final remarks: Keep your messaging simple; win as many people to support your way; so called Stakeholders are either Part of the Problem – or Part of the Solution; make it as easy as possible for people to change! Do whatever you can do to shape the world we want it to be for generations to come!

Buddy Brewer (@bbrewer) from SOASTA hitting on the problem of different meanings and definitions of Performance to different people. Also the metrics these people use. Key is to find the relationship between these metrics that matter to Ops, Dev, Test and Business.

There were more presentations that you should check out online as the keynote has been recorded and will be available on the conference site soon!

Stay updated