ScaleConf New Zealand 2018
6 - 7 Aug 2018 at Museum of New Zealand Te Papa Tongarewa

In this introductory workshop, we will get into practical examples of how to create a Progressive Web App -an offline-capable app-like web application- from scratch using the latest specs. We will cover desktop and mobile PWAs, using Service Workers and other APIs, debugging techniques and tools available. We will learn how to detect connection status and how to create a successful offline experience. We will finally see how to use Web App Manifest and other specs to let your users install your PWA on Android, iOS, and desktop OSs, how to use analytics and enhance your PWA with platform-specific APIs, and the techniques to publish it into app stores. After this workshop you will have a fully functional PWA that you can serve to your users.

In the workshop we will build InnoDB Cluster and see how it works in real life, with real web application, in two modes: Multi Master and Single Primary. Then we will build NDB Cluster across the participants machines and will show how it survives disaster and recovery and online scaling.

Considering deploying your containerized application to production? Whether you are deploying to bare metal or a cloud VM, container orchestration frameworks can reduce the operational load of managing, updating and scaling your application. Kubernetes is a mature orchestration platform with over a decade of development behind it. This workshop will be a hands-on introduction to Kubernetes. Learn how to set up a Kubernetes cluster, deploy a containerized application and manage updates with rolling upgrades. (3-4 hrs) Workshop Agenda: - Introduction Containerised Applications Kubernetes background - Bootstrapping Kubernetes Cluster - Deploy containerized application - Development Workflow Update application Rolling upgrade

The Cloud is changing the way we build and scale systems. But while the technology is mature, the challenge remains how we align this to our motivation and strategy. In this talk I will discuss the layers that live above the technology, starting with a cloud strategy and ending with who does what and why. Not understanding the dependencies between these layers is a risk and direct cost on every cloud project. By modelling the business patterns for cloud support, security and governance, organisation can make better decisions, act faster and scale more effectively.

TBC

Seems like all you hear about these days are Single Page Applications. Angular, React, Vue, Ember are transforming the way we think about the frontend. But what about securing these applications? This often tends to take a back seat to speed, animations and other cool features of these frameworks. Between cookies, tokens, keeping users authenticated, and handling resource access, securing these apps can be tricky - you may even feel like you need a second page (gasp!) for your authentication setup! But we have the technology to create truly secure single-page applications. In this talk, we’ll walk through securing an Angular application, but our approach will apply to nearly any single-page application framework.

Do you have services where the owners claim they run at five 9's but you often run into errors? It's very easy and convenient to build metrics at the service level. These often hide a wide array of issues that users might face. Having the right metrics is a key component of building reliable products. This talk goes into the design of these metrics, real world examples to illustrate good/bad designs.

In 2017 Heptio launched Contour, a Kubernetes Ingress controller built using the Envoy proxy project from Lyft. In this session I'll cover two unequal parts What is an ingress controller and why should you care, which leads into an overview of the design of Contour, why we chose Envoy, and so on. What can the audience, as potential authors of Kubernetes controllers (or operators, choose your poison) learn from our experiences developing Contour. The presentation gives real world advice on the traps involved in the components that make up a kubernetes controller; Go dependency management, handling concurrency issues, handling configuration, functional testing, and staying sane while living with docker.

When migrating to the cloud, certain tasks need to be evaluated in terms of managing and architecting your infrastructure and applications for optimal performance. In this session we’ll cover: Reliability Stability User Experience Scalability MTTD && MTTR Additionally, we’ll touch on a couple of options for full-stack instrumentation to maximize the benefits of cloud migration. Auto-Scaling && Capacity Management - Debunking Myths && Gotchyas Monitoring - Deviaton Detection - Event Correlation - Alerting

One certainty in the world of microservices is that these new architectures don’t reduce network and application traffic. In contrast, the volume of application traffic going over the virtual microservices wire grows exponentially with every new service that comes online. API and service control traffic over the control plane; scheduling traffic and policy routing over the management plane; ingress and container-to-container traffic over the data plane...the more agile our service delivery becomes, the more complexity we bring to data on the wire. In this presentation, we’ll discuss the current state of solutions available to manage, control, and deliver microservices application traffic without sacrificing the core tenets of application performance and availability.

Test automation suites often have - at least - three problems. Teams complain of flaky tests; tests take too long to run; and only a few specialists know what to do if a test fails. Daniel McClelland believes these roadblocks have something in common. All too often, test-focused infrastructure is as 'black-box' as the tests it runs. Daniel will outline how Trade Me has embraced 'observability', to overcome these hurdles. Monitoring, alerting, log aggregation, and dashboards all play a part. The most important solution though? To grow ownership of test automation out across the company. Doing this has helped Trade Me scale up its daily automated testing by 12,000%.

What might seem like small decisions have a habit of growing into complex problems. When Travis CI was started 6 years ago, it was as a free open source service focused on helping the community. After running a successful crowd funding campaign to add support for private accounts, we decided to keep travis-ci.org for open source projects, and use travis-ci.com for private projects. Little did we know that this would lead us down an interesting, and complex path, when we decided that we needed to merge these systems and customers experiences into one. Merging these two platforms together involved major changes to both our application and infrastructure stacks. With over 6 years of data, 500,000+ users, 1.3 million repositories, and terabytes of data, making sure our users experience was not disrupted was paramount. This is the story of how a product decision, which required major technical work, ultimately led to a better customer experience.

For the past 15 years, Google has been following a set of practices for keeping systems operational called Site Reliability Engineering (SRE). Recently, Google began speaking publicly about SRE, publishing a book, releasing new features as part of their cloud offering, and much more. At the same time, communities raised concerns about the conflation of DevOps and SRE. Is SRE "DevOps 2.0"? Is SRE trying to overtake DevOps? Is DevOps dead? Can I adopt DevOps and SRE at the same time? Come to this session to get answers to these questions and more :)

After we understand how important Web Performance is and we've applied basic techniques, what's next? In this session, we will cover extreme techniques that will blow your mind about Web Performance. New compression algorithms? new image formats? Client Hints? HTTP/2 Push? Let's hack Web Performance. After quickly reviewing current metrics and goals for web performance, we will get deeply into advanced hacks to increase perception and performance on the web and PWAs. We will cover the updated checklist of techniques to improve experience and perception and recent additions and hacks, such as: the usage of HTTP/2 push or not, QUIC, how to replace PNGs and JPEGs, HTTP Client Hints, better compression encoding algorithms available, browser's prefetching and lazy loading new abilities and how advanced techniques on top of Service Workers can help improving conversion and reduce user's frustration. We'll cover some extreme new ideas, such as serving ATF content in 14Kb, using magician tricks or using new APIs to do reactive performance optimizations. We will finally talk about how we always underestimate the Web on mobile devices and what we can learn from the AMP framework.

I would like to share our experience at FMG with how we use scripts and Jenkins to allow developers and testers to restore production database copies for testing easily. As well as the technical details I will discuss our motivation and the results for the business. The insurance software we use is very tightly linked to its databases. We regularly need to have up-to-date data to resolve client problems which means we need to restore production data to test environments frequently. Traditionally this task was done manually by our DBA. However, I am that one and only DBA, and this did not scale well. The demands from developers and testers on database resources are only increasing over time. In the past we were constrained by those resources, however with modern software and hardware this is less of an issue. Our constraints now are more the human resources needed to maintain the systems. Automating these tasks allows every developer and tester to clone their own copy of the production database, on demand. Aside from the scaling factor automating the process allows us to schedule restores during off peak time so the impact to the teams is minimised. A database restore job can take up to two hours and slows the database server down for everyone for the duration. Automating the database restores streamlines our testing process; provides up-to-date data for everyday support; and empowers developers and testers to take control of their environments. In this talk, I will cover the following topics: An overview of our architecture and environment at FMG. Technical details on how the scripts work. The Jenkins-based interface. The benefits it brings. I believe this talk could appeal to many IT professionals as more and more companies are adopting the CI/CD practices necessary for scale, and where automation is the key to success. In our case greatly improving the speed and efficiency of our testing cycle.

We built one of the largest platforms using serverless technologies like AWS Lambda (we don’t run a single server, anywhere). It has been an incredible journey and we learnt a lot of lessons in the process. In this session we’ll share notes from our adventure including: What our serverless system looks like 2 years in (yes, I’ll share stats). Key design patterns and architectures you will need to know to succeed. Common serverless mistakes and how to avoid them. Our top ticks and tricks for serverless success and happiness. We will dive in to the design of our platform and share interesting data, go through key patterns and architectures, and discuss what it actually takes to build a scalable, reliable and high-performing serverless system today.

You probably have dozens of applications, api's, websites that are running on various flavours of Windows Server and IIS - either on physical machines, or in virtual machines. You've heard about the advantages of containers - how you can spin them up anew in seconds, treat your apps like cattle - not pets, scale-out your app as demand requires. In this talk you will hear about Windows Server containers, see how you can auto-build them to deploy your app, even see a Classic ASP with ActiveX deploying and running. We'll also look at the upcoming changes in Windows Server that allow you to run both Windows and Linux containers on the same host!

The data heart of many businesses is monolithic legacy systems, beating powerfully but slowly as they are hardened against rapid change. However, to quickly derive business insight, act for customers in real time, and empower autonomous product teams, we look to microservice architectures and self-service data platforms to infuse the entire organisation with arterial streams of data. To make the change is equivalent to open-heart surgery - how do we maintain the data pulse while we animate every part of the business with responsive data? The good news for the patient is that we have surgically precise techniques available. Andrew will share a range of approaches to common scenarios when migrating incrementally from monolithic legacy systems and batch processes to modern streaming architectures and services bounded by business contexts. He will also share some general design principles for modern data architectures. This will provide a practical framework for thinly slicing data migrations that your organisation can start using today.