CWIN17 telford api management, practical implementation experience - david rutter

Editor's Notes

• #4: Back in 2000 Salesforce and eBay made their first APIs public Today 50% of Salesforce revenue is from APIs For eBay its 60% Expedia is of note because 90% of its revenue is from APIs. When you use Expedia or one of its brands you might think that is just a search engine for hotels. Its actually a complex marketplace where hotels can bid for placements. Whilst all of these companies have disrupted their industries, their APIs are sustaining innovations, by that I mean the APIs are enhancing what the companies already do. They are following Conways law.
• #6: Whilst all of these companies have disrupted their industries, their APIs are sustaining innovations, by that I mean the APIs are enhancing what the companies already do. They are following Conways law.
• #7: I wanted to give you an example closer to homeTag: is an advertising production company, in Clerkenwell London; you probably haven’t heard of it but you will have seen its work on billboards, in newspapers, and TV. Back in 2000 it was one of many production companies with a revenue of a few Million.When I left it had a turnover of 100Million, and a client list that an SI would dream of. Its success was down to several things, a business vision, a SAAS based platform for use by customers and suppliers, and finally integration with its customers and suppliers systems using APIs. In this case what Tag did was disruptive to its market in the UK and worldwide and also internally; it was a small business which was able to readily adapt to new business models and practices. The disruptive innovation typically comes from 3rd parties using your APIs in ways you haven’t thought about, or in ways that are not worth your while investing in. In the early days of Twitter, people typically tweeted using a third party application, Tweetdeck, which used the Twitter APIs. Tweedeck was so successful, Twitter had to buy it.
• #8: Jeff Bezos issued his Api mandate in 2002; Amazon is another great example of disruptive innovation, and its driven by its APIs. For example. the marketplace APIs allow you to see all the prices of all vendors; you can outsell Amazon on Amazon, all though its APIs.
• #9: We touched on the need for various levels of testing and devops, so with that in mind I will hand you over to my colleague Abhilash
• #10: Most of the API Management vendors will provide the same key features - A Developer Portal, to promote and catalog the APIs.- API Gateway, to enforce policies such as Access Control, API Transformations, Routing and Aggregation of back end APIs. -API Management Portal to provide API lifecycle management, approval processes for API registrations, user and API key management, and to manage the onboarding of APIs onto the API Gateway. The portal also provides service and support capabilities. - Analytics of API Usage and performance for both the service team and external developers.APIm is a ‘no code’ or ‘lowcode’ environment, so we need to integrate into our source control and automation environments. If we are using API first and shift left and shift right, we will need stubs, testing and monitoring tools Finally to provide the Channel specific APIs with BFF capabilities we will need some local processing and data storage, and perhaps notifications.So actually what you should be looking for is not just an APIm tool but an iPASS platform.In our case this was APIM from IBM on Bluemix, other projects have used Mulesoft on Amazon, and Capgemini has some iPAAS offerings.
• #11: Most of the API Management vendors will provide the same key features - A Developer Portal, to promote and catalog the APIs.- API Gateway, to enforce policies such as Access Control, API Transformations, Routing and Aggregation of back end APIs. -API Management Portal to provide API lifecycle management, approval processes for API registrations, user and API key management, and to manage the onboarding of APIs onto the API Gateway. The portal also provides service and support capabilities. - Analytics of API Usage and performance for both the service team and external developers.APIm is a ‘no code’ or ‘lowcode’ environment, so we need to integrate into our source control and automation environments. If we are using API first and shift left and shift right, we will need stubs, testing and monitoring tools Finally to provide the Channel specific APIs with BFF capabilities we will need some local processing and data storage, and perhaps notifications.So actually what you should be looking for is not just an APIm tool but an iPASS platform.In our case this was APIM from IBM on Bluemix, other projects have used Mulesoft on Amazon, and Capgemini has some iPAAS offerings.
• #12: In terms of vendors the were a number of pure play vendors concentrating on API management and some integration vendors that also had API Management capabilities.
• #13: Integration vendors realized that they were missing out so there has been some consolidation. If you want to understand more about how to select a vendor please talk to Lisa who covered this in her earlier talk.
• #14: For my current client, we created a digital viewpoint of where they needed to go,and API management was one of the features. This infographic was one of the outputs from some follow up sessionsto identify which business services should be externalized and most importantly who the stakeholders were.
• #15: API Identification DDD, IAF, TOGAF BFF, Business, SoR APIs SOA vs Microservices FAAS
• #16: We touched on the need for various levels of testing and devops, so with that in mind I will hand you over to my colleague Abhilash
• #17: The following slides show the techniques we have used on our project to ensure testing is Part of Rapid Delivery and not a blocker, like it is on many projects.
• #18: The first tool in the toolbox is ‘API Frist’.The term ‘API First’ is ambiguous.What we mean by ‘API First’ is that we design the API at the start of the project. 
• #19: So on a recent project we designed the API First, and defined it in Swagger.This meant that the consumers and producers of the API could work in parallel,knowing in advance that the API would deliver what everyone needed. It also meant that we could create stubs, in advance of the real API being availableand also that the testers could start their implementation
• #20: TDD is normally accepted as best practice for developers – so that they can develop unit tests. Shit Left: This follows the maxim "Test early and often." If we take that to the next level, if I have a stub, I can develop my scenario tests in parallel with the developer. This verifies that stub API for consumers and we can also apply it to the real API during development.
• #21: Once we have shift left, we can build it into our CI framework.With our current project, shift left testing is triggered manually
• #22: This is how we would like our Shit Left testing to work.This is something I have prototyped, and have shown to Colleagues in some internal demonstrations.
• #23: If our project we take the same tests that we use for development RCA and use them in production. External monitoring should be more than availability testing, and include functional, security and performance, to provide detailed alerting of issues.
• #24: In an API environment, it is critical for developers and later for Ops teams, to be able to quickly diagnose the root causes. Shift left allows us to do this for developers Note: This image requires artworks SOURCE https://docs.axway.com/u/documentation/api_gateway/7.5.2/ Axway- Configure API Gateway high availability
• #25: So we take the tests for development and run them using alert site.This is integrated into the clients ServiceNow incident management system. The tests allow us to identify not just that there is an issue but also the likely component https://polymath.com/flat-faces-icons-circle-6/ https://www.iconfinder.com/icons/416400/business_customer_service_support_icon#size=128
• #26: This slide illustrates the tools and techniques that we are using on our current project,which we expect to go live in a weeks time. ICON SET from: http://wararat-keotkuea.blogspot.in/ https://pngtree.com/so/Businessman-0-1-0-1 https://www.shareicon.net/kid-user-child-avatar-girl-female-young-103156
• #28: The first lesson is that APIs require business change The APIs need a business stakeholder We need governance around the APIs and change management The Operations team may need additional training and support Designing good APIs is not easy and many development teams, are inexperienced in creating, testing and using them,a competency centre approach may be needed
• #29: API first allows teams to work in parallel All the teams we worked with were using an agile approach, There needs to be architecture at the beginning and ongoing to ensure the dots join up Many of the tools allow no- or low code approaches – we still need the same level of QA
• #30: Testing early and continuously saves time; testing becomes an enabler of QA, not a hindrance.Test right ensures when we go into production are ops team are better informed, better supported
• #31: The aim of this presentation was to give you a viewpoint on the importance of APIs to modern business, and an insight into what it involved, based on a real life project.