Prototyping your Next Big Idea - An introduction to Google Cloud Platform
- 1. Prototyping your Next Big Idea: An Introduction to Google Cloud Platform Xin Ying Lim
- 2. What stops you from getting your next big idea off the ground? No Time No resources (eg financial backing) Not sure how do I get started
- 3. Google Cloud Platform (GCP) A suite of cloud computing services. 150 services, including: Website/server hosting: App Engine, Compute Engine, Kubernetes Engine Database: Firebase Realtime Database, Firestore, BigTable, Cloud SQL Data Analytics: BigQuery, Dataflow, Dataproc (managed Spark and Hadoop instances), Functions, PubSub AI and Machine Learning tools: Cloud ML Engine (managed Tensorflow), Text-to-speech API, Cloud Video Intelligence, AutoML
- 4. An Idea: Using Weather Data to configure IOT devices in a restaurant
- 5. The idea
- 6. Break into three steps Weather API data Data Processing IOT devices Sent for Send configuration changes to Website that allows users to choose a weather, then send weather data 1 2 3
- 7. Step 1: Building a web app
- 8. How the app works
- 9. Database: Cloud Firestore Flexible and scalable NoSQL database Real-time listeners: Allows listening for changes in data
- 10. Hosting the website: Firebase Hosting Run firebase deploy, and that’s it It will: ● Give you a free domain ● Configure SSL for you
- 11. So far.. Getting weather data Website (normally a weather API) Updates current weather Database Sends weather data ? Updates database
- 12. Step 2: Building an API to handle weather changes
- 13. {“weather”: “sunny”} ? {“fan”:”on”, “umbrella” : “open” “menu” : “menu_for_sunny_days”} What goes in here?
- 14. Cloud Functions Serverless, scalable functions Used to connect services together Can be triggered via HTTP Just pass it a function and it will take care of provisioning an instance to run the function on
- 15. Use Cloud Functions to convert Weather data to Device Configuration data Fan Umbrella Menu Recommendation Sunny On Open Sunny Cloudy Off Close Cloudy Rainy On Open Rainy {“weather” : “sunny”} is converted to {“fan” : “on”, “umbrella” : open”, “menu” : “sunny”} Devices Weather
- 16. So far.. Getting weather data Website (normally a weather API) Updates current weather Database Sends weather data Updates database Data Processing ? Cloud Functions Sends New Device Configuration
- 17. Step 3: Build the hardware, connect them to the cloud
- 18. Things I used: 1. Motors and servos 2. Raspberry Pi: a miniature computer a. Its GPIO pins allow it to control devices eg. Motors 3. Cloud IoT Core - a service to connect devices into Google Cloud Platform Cloud IoT Core Sends configuration changes to controls Servos and motors
- 19. Cloud IoT Core Use it to manage fleets of IoT devices and handle authentication Authentication: Private/public key pairs and JSON Web tokens Connects to IoT devices through MQTT or HTTP protocol
- 20. Cloud IoT Core: Sending and Receiving Data It does two main things: 1. Send data (telemetry, state data) 2. Receive data (configuration changes, one off commands) In the case of telemetry data, data sent to Cloud Pubsub For configuration changes, data is sent to devices via MQTT connection
- 21. Cloud IoT Core Client on Raspberry Pi listens for configuration changes over MQTT connection On change in configuration, the Raspberry Pi will configure the devices accordingly Devices Sunny Cloudy Rainy Fan On On, at lower speed Off Umbrella Open Close Open Menu Sunny Cloudy Rainy
- 22. Cloud IoT Core Sends configuration changes to controls Servos and motors Configuring Iot Devices Getting weather data Website (normally a weather API) Updates current weather Database Sends weather data Updates database Data Processing Sends New Device Configuration Cloud Functions The full project
- 23. The best part: Scalability
- 24. Ways to get started! Documentation Codelabs Cloud.google.com/blog QWIKLABS (follow their twitter for info on free credits) Free $300 USD Credits Google Cloud Platform podcast For startups: apply for $3000 in credits
- 25. Thanks! Feedback Form How to get started with GCP
Editor's Notes
- #2: XY, Software Developer, GDG Meetup
- #3: There’s an easy solution to the two is to start small. Build a working prototype first as a small project. Then scale up. Today I will be telling you how you could get started, by introducing the basics of Google Cloud Platform
- #4: Different services covering all you need to build working products. Your everyday apps are built on the same backend system GCP has, YouTube, Gmail, drive 150 services, can be very overwhelming to get started. So, i’ll use the project I made for today’s showcase as an example, introduce you to the services I used. First, we come up with an idea
- #5: Okay okay, hear me out: Before you start pitching then
- #6: I will show you an example of a prototype of a project. Getting the weather data:
- #7: For the purpose of the event, we will have a webpage Backend figures out what devices do what when it’s sunny. Maps weather patterns to devices
- #8: Normally you would call an api to get the current weather and use that to configure your device, but weather variation do not change interestingly within the span of minutes. For the purposes of this demo , we have a web app to choose (
- #9: It is currently ___ (current weather) ___
- #11: Google has many options for website hosting. App engine, container engine
- #12: What do we do with the weather data?
- #14: When you receive data from the API/web app need to convert it to instructions for devices. What does this?
- #15: Give it a function at it will run forever, spinning up instances only when needed You can trigger it in one of three ways:
- #17: What do we do with the weather data?
- #19: Motors and servos the actual devices moving the umbrella, turn on fan etc The raspberry pi is
- #20: Connecting traditionally unintelligent devices to the cloud. MQTT is a lightweight messaging protocol optimized for unreliable networks It is designed for connections with remote locations where a "small code footprint" is required or the network bandwidth is limited A message broker and two types of clients. Receives all messages from clients. Message broker and a number of cients
- #22: Devices are listening for configuration changes over the MQTT connection
- #24: Selling to restaurants everywhere a smart iot system Cloud iot configurations Cloud functions can do how many invocations. Cloud firestore can store data Fire base hosting: auto scaling You are building on the shoulders on giants. Leverage the infrastructure they can give you to push forward your big plans
- #25: Okay, you’re interested, now how to get started? QWIKLABS Nothing beats actually doing its free credits If you’re a startup, DevFest you can apply up to 3000 credits Blog contains case studies of what other companies have done