GCP 101 Products and Services

#gcpug #sendai #1
雑な資料でごめんなさい・・・

Transcript

1. GCP 101 - Products and Services - @sakajunquality 19.03.26 #gcpug #sendai
2. About me - @sakajunquality - Google Developers Expert, Cloud - Software Engineer, Reliability @ Ubie, Inc.
3. Agenda - Cloud Computing - GCP Products - Getting Started
4. 1 Cloud Computing
5. Cloud Computing - Networking - Compute - Database and Storage - Big Data and Machine Learning - Developer Tools
6. 2 GCP Products
7. GCP Products - Networking - Compute - Database and Storage - Big Data and Machine Learning - Developer Tools
8. GCP Products - Networking - Compute - Database and Storage - Big Data and Machine Learning - Developer Tools
9. Network
10. Network - VPC VPC - Virtual version of a physical network - RFC1918 Private Network - IPv4 unicast - Network and Subnet
11. Network - Load Balancing Cloud Load Balancing / Cloud CDN / Cloud Armor - Anycast IP Global Load Balancer - From zero to full-throttle - CDN and WAF
12. Network - VPN / Routing Cloud Nat / Cloud Router / Cloud VPN / Cloud Interconnect - Managed NAT service - Managed Router - Managed VPN Endpoint - Fixed Line Endpoint
13. Network - VPN / Routing Cloud Nat / Cloud Router / Cloud VPN / Cloud Interconnect - Managed NAT service - Managed Router - Managed VPN Endpoint - Fixed Line Endpoint
14. GCP Products - Networking - Compute - Database and Storage - Big Data and Machine Learning - Developer Tools
15. Compute
16. Compute Types - Virtual Machine - Container - Serveless
17. Compute Types - Virtual Machine - Container - Serveless
18. Virtual Machine - Compute Engine Compute Engine - IaaS - Scalable, High-Performance VM - Live migration - Automatic Discounts - Preemptible Option
19. Compute Types - Virtual Machine - Container - Serveless
20. Container - Kubernetes Engine Kubernetes Engine - Full-managed Kubernetes - Deploy “Containered” Application
21. What’s Kubernetes? - Container Platform based on Google’s Borg - Borg has supported Google’s service over 12 years - Orchestrates computing, networking, and storage infrastructure - Microservices Platform - OSS written in Go - Borg is written in C++ - “Graduated” CNCF project - Crossed the Chasm https://speakerdeck.com/sakajunquality/starting-google-kubernetes-engine-2019
22. Compute Types - Virtual Machine - Container - Serveless
23. Serverless - AppEngine AppEngine - Full Managed Application Platform - Triggered by HTTP(S) - Standard and Flexible
24. Serverless - Cloud Functions Cloud Functions - Function as a Service - Triggered by Events and HTTP
25. Which Compute Should I use?
26. Which Compute Should I use? Up To You
27. - Workload Type - HTTP triggered or Event triggered or cron triggers? - Languages - GAE Supported? - Monolithic? Microservices? Considerations for Compute Resource
28. GCP Products - Networking - Compute - Database and Storage - Big Data and Machine Learning - Developer Tools
29. Database and Storage
30. Database - Cloud SQL Cloud SQL - Managed MySQL and PostgreSQL - Private IP Access - Auto-Backup
31. Database - Cloud Datastore Cloud Datastore - Full-managed NoSQL Database - Automatic Sharding and Replication - ACID transactions - SQL-like queries - Indexes
32. Database - Cloud Firestore Cloud Firestore - Full-managed NoSQL Database - Next Generation of Cloud Datastore - Datastore mode - Realtime Database - ACID Transaction - Multi-region replication
33. Database - Cloud Bigtable Cloud BigTable - Low latency, massively scalable NoSQL - Eventually Consistent
34. Database - Cloud Spanner Cloud Spanner - No-Compromise Relational Database Service - Global Scalable and strongly consistent
35. Database - Cloud Memorystore Cloud Memorystore - Full managed Redis - Redis cluster is not supported yet
36. Which Database Should I use?
37. Which Database Should I use? Up To You
38. Considerations for database - Use traditional RDBMS? - CAP Theorem - Datasize? - Scalable?
39. Storage - Persistent Disk Persistent Disk - Block storage - can be mounted on GCE
40. Storage - Cloud Storage Cloud Storage - Object Storage
41. GCP Products - Networking - Compute - Database and Storage - Big Data and Machine Learning - Developer Tools
42. Big Data
43. Big Data - BigQuery BigQuery - Serverless - Highly Scalable - Full-Managed Data Warehouse - Standard SQL
44. Big Data - Cloud Pub/Sub Cloud Pub/Sub - Event Stream (Queue) - Data Analytics - Event-driven Computing
45. Big Data - Cloud Dataflow Cloud Dataflow - Managed Apache Beam - Batch + Stream data processing
46. Big Data - Cloud Composer Cloud Composer - Managed Apache Airflow
47. Example Use-case of Big Data Product
48. Machine Learning
49. Machine Learning - ML Engine ML Engine - Create Custom Model - Tensorflow / XGBoost / scikit-learn - Managed JupyterLab with Deep Learning Image
50. Machine Learning - BQ ML BigQuery ML - Machine Learning with BigQuery - Linear regression - Binary logistic regression - Multiclass logistic regression for classification
51. Machine Learning - AutoML AutoML - Create Custom Model Easily - Just Prepare dataset - Vision - Natural Language - Translation
52. Machine Learning - ML APIs ML APIs - Trained Model by Google - Available as API
53. Which ML Product Should I use?
54. Which ML Product Should I use? Up To You
55. If you are a data analyst or marketing… - Try BQ ML Suggestions for ML on GCP
56. If you are a software engineer… - Try pre-trained model with ML APIs first - Try custom model with AutoML - Consider custom model with ML Engine Suggestions for ML on GCP
57. GCP Products - Networking - Compute - Database and Storage - Big Data and Machine Learning - Developer Tools
58. Developer Tools
59. Developer Tools - Cloud Shell - Cloud Build - Cloud Source Code Repository - Cloud Registry
60. GKE Ecosystems: CI/CD Cloud Build - Fully-managed CI - 120 min free tiers per Day - Native Docker Support - Custom Builders - https://github.com/GoogleCloudPlatform /cloud-builders - https://github.com/GoogleCloudPlatform /cloud-builders-community https://speakerdeck.com/sakajunquality/starting-google-kubernetes-engine-2019
61. GKE Ecosystems: CI/CD More Kubernetes Specific: Simple GitOps from GitOps-style continuous delivery with Cloud Build https://cloud.google.com/kubernetes-engine/docs/tutorials/gitops-cloud-build?hl=en
62. Monitoring
63. Monitoring - Stackdriver
64. GKE Ecosystems: Monitoring Stackdriver Monitoring - Full-Stack Monitoring Suite - New Features - New Kubernetes Dashboard - Istio Integration - Prometheus Integration https://speakerdeck.com/sakajunquality/starting-google-kubernetes-engine-2019
65. GKE Ecosystems: Monitoring Stackdriver Logging - Full-managed Log platform - GKE has pre-installed fluentd - Collects Stdout/Stderr of containers - Parse JSON to structured log https://speakerdeck.com/sakajunquality/starting-google-kubernetes-engine-2019
66. 3 Getting Started
67. Getting Started - Create a new account - Try “Quickstart” - Official Support - Community Support
68. Create a new account https://goo.gl/2ib1Hs
69. Try “Quickstart”
70. Official Support - Basic - Role-base - Enterprise - https://cloud.google.com/support/
71. gcpug.jp Community Support
72. 4 Live Demo
73. Live Demo Maybe
74. 5 Takeaways
75. Takeaways - Plenty of product is available in GCP - Try it out with “Quickstart” - Enjoy #gcpug