December 2022 Summaries

Firebase Cloud Messaging (FCM) is a free service for sending notifications across Android, iOS, and web applications, with targeting tools designed to reduce irrelevant messaging and improve user engagement. Through Google Analytics, developers can target users using default properties such as location, app version, language, first open date, and recent engagement, or define custom properties like hiking skill level and region. These properties can be combined into reusable audiences for more precise campaigns, while Firebase Console provides notification-open and conversion metrics to evaluate results. Because Analytics properties have quotas and generally support only one value per property, FCM topics allow users to subscribe to multiple interests, such as individual trail-condition updates, with FCM managing membership and message delivery. For highly customized groups identified through server-side analytics, imported segments use BigQuery and FCM registration tokens to target selected users. Together, these methods enable developers to personalize campaigns, manage delivery infrastructure, and measure notification effectiveness.

The tenth and final Firebaser Friday profile introduces Marina Coelho, a newer Firebase team member whose work includes Firebase blog and YouTube content. Marina is developing a series of articles on creating Android applications with Jetpack Compose alongside Firebase services such as Firestore, Crashlytics, and Firebase Authentication, while also studying iOS development. She cites receiving a plush Sparky mascot as a favorite Firebase memory, enjoys music from Bring Me The Horizon as well as artists including Ed Sheeran, Frank Ocean, and John Mayer, and recommends rereading The Little Prince for its reflections on love and respect. The post also directs readers to her Twitter account and highlights Firebaser Friday as a monthly series featuring Firebase team members and their work.

Firebase can be integrated with Google BigQuery to support custom, large-scale analysis of app and user data beyond the standard Firebase console dashboards, including combining Firebase data with other sources and retaining it for longer-term use. BigQuery exports are available for Analytics, Cloud Messaging, Crashlytics, Performance Monitoring, and A/B Testing, while a Firebase Extension can export Cloud Firestore collections; setup requires appropriate IAM permissions and configuration through the Firebase console’s Integrations page. Although standard exports generally make the previous day’s data available, Google Analytics and Crashlytics offer streaming options for current-day or real-time analysis, such as identifying purchase intent quickly or powering live crash dashboards and alerts. Analytics streaming uses temporary intraday event tables, may omit some delayed or failed data, has no historical backfill before linking, and standard properties face a daily one-million-event export limit unless upgraded to Google Analytics 360. Crashlytics provides both daily batch tables, including up to 90 days of backfilled data for trend analysis, and real-time tables without backfill, which can be combined in queries. BigQuery includes free capabilities but charges for storage, streaming inserts, and queries, while its Sandbox permits evaluation without payment details but imposes limits such as 60-day table expiration and no streaming access.

Firestore introduced preview features aimed at improving scalability, performance, and database management: count aggregation queries, reduced scaling limits, and time-to-live policies. Count queries, available in the Web, Android, iOS, and Node.js SDKs, return the number of documents matching standard or collection group queries without requiring clients to download those documents, though they are currently server-only and lack offline or real-time support. Firestore also removed previous maximum limits on reads per second, writes per second, and concurrent real-time connections, reducing the need for high-scale applications to design around such constraints. TTL policies allow developers to designate a timestamp field that triggers automatic document deletion, typically within 72 hours after expiration, helping manage growing datasets and compliance-related retention requirements.

Firebase serverless products integrate with Google Cloud Monitoring to improve security observability, particularly for Firebase Security Rules and App Check. Security Rules metrics, including access denials, are available in both the Firebase Console and Cloud Monitoring’s Metrics Explorer, where teams can filter and analyze rule evaluations in detail. Teams can create alerting policies that notify support staff when denied requests exceed a chosen threshold, such as 20 queries per second for five minutes, while configuring notification channels, severity labels, incident-response documentation, and handling for missing data. Such alerts can reveal faulty rule deployments or potential malicious access attempts and automatically close when denial rates return to normal. App Check metrics can similarly support alerts for spikes in invalid Firestore requests or declining numbers of outdated clients, helping teams identify attacks and determine when enforcement is safe. The guidance recommends using Cloud Monitoring as a baseline for Firebase production applications and extending monitoring to Cloud Functions and Firebase Authentication for performance, usage, and failed-login issues.

Firebase’s fifth-year Cloud Functions update introduces preview-stage second-generation capabilities alongside improvements for first-generation functions, emphasizing performance, deployment efficiency, and safer configuration. Second-generation functions support configurable concurrency, allowing each instance to process multiple requests and reducing cold starts during traffic spikes, while also adding triggers for Authentication blocking functions, Firebase Alerts, custom extension events, and Eventarc integration that can trigger Cloud Run services. New codebases let developers organize functions into isolated groups with separate dependencies and deploy only relevant components, while automated deploy skipping uses hashes of source files, secrets, and environment variables to avoid redeploying unchanged codebases. Parameterized configuration provides typed deployment-time variables for both generations, and secret parameters integrate with Cloud Secret Manager to protect sensitive values such as API keys. These features can be tested locally through the Firebase Emulator Suite, with second-generation and first-generation functions able to coexist in the same project and deploy through the Firebase CLI.