Firebase

Introduction

Firebase is a cloud based development platform provided by Google. Firebase was created by James Tamplin and Andrew Lee at Envolve in 2011 to funnel data storage and data transfer into one realtime system for the companys chat service. In 2012 Firebase goes public as a separate company to Envolve, and in 2014 Firebase is acquired by Google. The short version of Firebase is that it provides a complete Software Development Kit, or SDK, across multible platforms to create applications with, providing cloud based solutions such as storage, hosting, and serverless computing. This allows developers to focus on building applications without having to worry about managing and maintaining servers or scaling.

Brief History

2011: Firebase is created to funnel data storage and data transfer into one realtime system for Envolves’ chat service.
2012: Firebase goes public with Realtime Database as their first product.
2014: Firebase launches two new products: Firebase Authentication and Firebase Hosting.
2014: Firebase is acquired by Google.
2016: Firebase expands their product portfolio with Firebase Analytics, Cloud Messaging, Test Lab and Crash Reporting, Notifications, Dynamic Links, Invites, App Indexing, AdWords and AdMob, introduces a new console, new SDK, and moves their documentation.
2017: Firebase launches Cloud Firestore, a realtime document database that can synchronize data across platforms.

Features

The Firebase portfolio of products is divided into three categories: build, release & monitor, and engage. Each category has their own suite of products, tools, and integrations. To best illustrate the size of Firebases’ portfolio, I will discuss and elaborate on their products and services by category.

Build

Cloud Firestore is a NoSQL document database and is the next evolution of Realtime Database. Data is stored on documents and is grouped in collections. Becuase data is stored in collections rather than tables and rows like SQL databases, a database can scale horizontally rather than veritcally. Scaling in Firestore is automatic and is limited to 1 million concurrent connections and 10 thousand writes per second. Data is stored across multiple data centers regionally and multi-regionally which increases the reliability and performance of applications. Essentially, Cloud Firestore is constructed to solve performance issues with scalability while also reducing the complexity of data models and querying that Realtime Database has.

Realtime Database is a NoSQL database and is the original product Firebase procued since its inception in 2011. Data is stored as JSON and is synchronized across platforms and can be accessed from the client-side SDK, essentially being like an API. Data is cached locally and can be accessed in realtime, even offline, to improve the end-user experience.

Cloud Functions is a backend framework for Firebase. Much like express.js, it is a framework to write the logic to handle HTTP requests on the application. Cloud functions are stored in a Cloud Storage Bucket and the code is handled by Google’s servers, building the function source, creating new instances to handle multiple events.

Authentication is an SDK that leverages industry standard authentication methods like OAuth 2.0 and OpenID Connect. Firebase provides two primary methods of authentication. The first is FirebaseUI Auth and is a complete drop-in product, providing a standardized log-in UI that can be configured accept third party authentication, such as Google Sign-In, Facebook Login, GitHub, etc. The second is the Firebase SDK Authentication which is a suite of authentication methods such as email and password, phone number, and third party authentication. It works by passing the user credentials or third party OAuth token through the Firebase Authentication SDK. Then the credentials are then verified on Firebase’s backend services and returns a response. The authenticated response is session based, meaning the user can end and restart the application and continue their session without having to log in again.

Cloud Messaging is a messaging and notification service that uses the Cloud Functions environment to push messages to the end users. Messages can be constructed in the Firebase Console GUI, and users can be targeted granularly, by groups, or by categories through serverside code. As with other Firebase products and services, this architeture is focused on scalability.

Hosting is a hosting service that serves content over an SSL connection on edge servers on a global CDN.

Firebase ML is a mobile SDK that provides machine learning models with APIs in the cloud or on-device. The cloud or on-device API Firebase provides pre-trained models for text recognition, image labeling, object detection and tracking, face detection and contour tracing, barcode scanning, language identification, translation, and smart reply.

Release & Monitor

Crashlytics is a realtime crash reporter for mobile apps. It condenses performance and crash information into managable lists and provides recommended solutions.

Performance Monitoring is an SDK that collects data about an applications performance. It monitors things like an applications start-up time and HTTP network requests, and reports this information in the Firebase Console.

Test Lab is a cloud based end-to-end testing environment, or lab, to test mobile applications on different devices. An application is hooked onto a device through an API for a device running in a Google Data Center, allowing the application to be tested on the hardware it would be running on in a live environment.

App Distribution is another type of testing tool for mobile applications. Uploading IPA (iOS App Store Package) or APK (Android Application Package) files to a Firebase project allows Firebase to distrube the application to testers. Testers are invited through the Firebase Console and applications can be tested at different builds.

Engage

Google Analytics is an SDK that monitors and reports user behavior. Conditions can be configured for measuring any event trigger, views, demographic or specific details about eCommerce or ad revenue. It is one of the free services of Firebase and has unlimited number of reports available to the user. Data is collected and is reported and displayed on a dashboard in the console.

Remote Config is a cloud service that lets developers make changes to an application without having the user update the application. These changes can be isolated from the rest of the user population, effecting changes to specific users or groups of users. Remote Config allows developers to make changes in the Firebase Console and roll out said changes to the user population without having to create a new build. Remote Config can also be integrated with A/B Testing.

A/B Testing is tool to test any metric of an application in an experiment. It runs on top of Analytics, Cloud Messaging, and Remote Config, allowing you to test for any metric at different builds or variation to an application in sample pool of users. A/B Testing is powered by Google Optimize, which performs Bayesian statistical analysis of the experiment and generates a report of Observed Data and Modeled Data. Observed being totals, averages, and percentile differences, and Modeled being probability, percentile differences from baseline, and metric ranges.

Strength

Firebase has a comprehensive ecosystem, providing the all the tools developers would require to build, test, monitor, and distribute applications.
Firebase supports cross-platform development. Meaning that developers can develop applications for android, iOS, and web from an extensive list of programing languages and libraries, ranging from javascript, c++, and unity to react native, flutter and vue.
Focused on scalability. The structure of Firestore’s NoSQL document database and other products like Cloud Functions and Cloud Messaging offer safe horizontal scaling of any application.
Has a huge community of users and plenty of resources to learn from and engage with. Ranging from comprehensive documentation with codelabs and guides, to blogs, Firebase Summit, and social media.

Weaknesses

Vendor lock-in prevents product owners or developers to move to a different backend provider.
Offers only one type of database. Though the arguments for NoSQL document database or Realtime Database are strong, there are usecases for SQL databases as well.
Complicated pricing plan. Though their Spark plan is free, the pricing plan is automatically updated to Blaze after a certain threshold. And finding out how much you can expect to pay can be hard to predict, because it is entirely based upon the number of users you have.

Comparison

Although a case could be made for comparing AWS Amplify and Firebase, it seems much more natural to compare Firebase to Supabase since Supabase advertises itself as the open-source alternative to Firebase.

Supabase is the open-source “alternative” to Firebase, meaning that the technologies Supabase employs in their services are also open-sourced. Where Firebase’s infrastructure is proprietary, such as Cloud Firestore, Supabase uses the open-sourced database PostgreSQL. Because Supabase only employs open-source technology, there is no vendor lock-in and because the open source community is huge, there will always be amazing support for this product. This also means that the pricing plan is less rigid and cheaper than Firebase. So choose why Firebase at all if Supabase is cheaper, open source, and offers great support? We have to return to the idea of Firebase being an ecosystem. Supabase’s product portfolio is limited to 5 products, being: Database, Storage, Edge Functions, Authentication, and Realtime. Essentially, what Supabase offers is baseline to what Firebase offers. This is not necessarily in Firebase’s favor as Supabase projects will have more freedom to integrate products of their choosing. However, Firebase offers a complete package that caters to every need of building, testing, deploying, and monitoring an application.

Summary

Firebase is, at its core, a Backend as a Service type of product. But when compared to a service like Supabase, it becomes apparent that what Firebase offers is an ecosystem. At the risk of complete vendor lock-in and potentially complex pricing plan, Firebase offers a tool or solution to most problems that is built to scale incredibly well.

Firebase products are divided into three categories: build, release and monitor, and engage. In build we find the backbone of Firebase since its inception, the Realtime Database, and its evolution, Cloud Firestore, and the tools we would expect from a BaaS, like Authentication, Storage, and a framework to write HTTP endpoints. But here Firebase also offers an Emulator Suite where we can test and build locally, as well as Security Rules for accessing your database, and Machine Learning tools to integrate with your applications.

Although build offers amazing infrastructure for developers to build applications on, release & monitor and engage takes this further by providing SDKs, extensions, and integrations that allow developers to monitor and analyze their applications performance and end-to-end testing, as well as taking user testing to another level by being able to isolate users with Remote Config and deploy changes to these specific users without having to update the entire application.

When choosing to employ a BaaS, there are many things consider. And although there are many great things that Firebase offer, choosing a BaaS can be case specfic. In some cases a relational database like what Supabase offers would be a better solution than the non-relational databases that Firebase offers. That being said, few offer an entire ecosystem like Firebase.

Credits

Bjørn Arne Læknes (barlaek)