GCP¶
GCP - Google Cloud Platform. Sometimes, it is hard to find out what Google uses behind the marketing names. Here is a quick overview of my review on it. It may not be complete or may contain errors, but it helps me if someone comes up with ‘Dataflow’.
Compute Services¶
Compute Engine – Virtual machines (VMs) powered by KVM and Linux-based hypervisors.
Kubernetes Engine (GKE) – Managed Kubernetes (K8s) clusters using containerd runtime.
App Engine – Serverless PaaS based on Google Borg (the precursor to Kubernetes).
Cloud Functions – FaaS (Function-as-a-Service) built on Node.js, Python, Go, Java, and .NET runtime environments.
Storage & Databases¶
Cloud Storage – Distributed object storage using Colossus, GCP’s next-gen file system.
BigQuery – Serverless data warehouse powered by Dremel (Google’s columnar storage and query engine).
Cloud SQL – Managed PostgreSQL, MySQL, SQL Server using Google’s Spanner replication & HA techniques.
Firestore & Datastore – NoSQL databases using Firebase technology and Spanner for ACID compliance.
Cloud Spanner – Global-scale relational database using TrueTime API for strong consistency.
Networking¶
Cloud Load Balancing – Uses Maglev (Google’s high-performance load balancing tech).
VPC (Virtual Private Cloud) – Built on Andromeda, Google’s software-defined networking (SDN) stack.
Cloud CDN – Uses Google’s Edge Network and QUIC protocol for low-latency content delivery.
AI & Machine Learning¶
Vertex AI – AI/ML workflow orchestration using TensorFlow, PyTorch, and JAX.
Cloud AI APIs – Pre-trained ML models built on TPUs (Tensor Processing Units).
AutoML – Low-code ML based on Neural Architecture Search (NAS) and TPUs.
Security & Identity¶
Cloud IAM – Built on Google’s internal authorization framework (used in G-Suite).
Secret Manager – Encrypted storage using Google’s KMS (Key Management Service).
Security Command Center – Uses Google Chronicle security analytics and SIEM tools.
DevOps & Monitoring¶
Cloud Build – CI/CD using Buildpacks, Docker, and Kaniko for container builds.
Operations Suite (formerly Stackdriver) – Monitoring/logging powered by OpenTelemetry & Prometheus.
Artifact Registry – Secure container and package storage supporting Docker, npm, Maven.
Analytics & Data Processing¶
Dataflow – Serverless data processing using Apache Beam on Flink or Spark.
Dataproc – Managed Hadoop, Spark, Presto, Hive, and HBase clusters.
Pub/Sub – Real-time messaging built on Chubby lock service and Paxos consensus algorithm. (Is often compared to Kafka. While it is not built on Kafka, it serves a similar purpose for real-time messaging and event-driven architectures.)
Now: 2024/01/10