b20c:c011:5:b109::/64

	Features	Size	Latency	Comparable
Cloud SQL	Managed MySQL, PostgreSQL and SQL Server	Up to 30TB
Cloud Spanner	Unlimited scale relational database	PB	Regional Low Latency R/W Multiregional Low Latency Read	Oracle DynamoDB
Cloud Bigtable	No-SQL, Key Value, Low Latency	PB	< 10ms	HBase Cassandra
Firestore (Datastore)	No-SQL, document based, key:value pairs	PB	>10ms
Firebase
Memory Store	In-memory, managed Redis and Memcached compatible	Reids:300GB Memcached:5TB	<10ms

Cloud SQL Notes and Best Practices

• Encrypted and have automatic backup available.
• Data replication and automatic failover between zones

Cloud Spanner Notes and Best Practices

• Globally replicated ensures very high availability (99.999%)
• Strong consistency across all replicas as well as external consistency (commit order is preserved)
• Local availability for better response times
• Regional instances replicate across 3 zones.
• Multi-Regional instances replicate across multiple regions. This allows for read latency to be maintained but write latency is impacted due to the mechanisms to ensure consistency.
• Place read-write workloads with read-write replicas, read-only workloads with read-only replicas

Bigtable Notes and Best Practices

• Single key based index.
• Scale linearly with additional nodes
• Highly durable using Colossus magic

FireStore (Datastore) Notes and Best Practices

• Do not write to a document more than 1x per second
• Regional and multi-regional options are available, select based on costs, availability and latency
• Multi-regional availability is 99.999%
• Datastore mode allows for backwards compatibility to Datastore and is required for App Engine integration
• Supports offline data persistence

MemoryStore Notes and Best Practices

• Replication across zones for standard tier to provide HA
• Must use Private Service Access to connect.
• Memcached is in BETA.

Summary of use cases

	Use Cases
Cloud SQL	SQL Support for OLTP ERP, CRM, E-Commerce, Web
Cloud Spanner	SQL Support for OLTP Global Financial Ledger Supply Chain/Inventory Management
BigTable	Streaming time series data (IoT) Fraud detection Personalization Ad tech Financial Analysis
Firestore	User Profiles User Sessions State Management
Memorystore	Low latency in memory data storage

Organizations form the root of the resource hierarchy. Folders, Projects and ultimately resources flow up into an Organization.

• Cloud Identity is an Identity Service which manages users and groups. Cloud Identity can be configured to support federation with AD or AAD for users and groups.
• IAM manages the role based access to resources. IAM policies let you control who, has what type of access to which resource.
• The resource hierarchy provides an inheritance model to allow for structured application of IAM policies to resources.
• Organization Policy manages the actions which are allowed on a resource regardless of who is performing them. It provides for a centralized and programatic control over cloud resources within an Organization.

Policies are inherited from higher level resources.

• You cannot revoke a policy which has been inherited.
• Policies are not retroactive. Resources which are already in place but violating a policy must be manually remediated.
• Moving a project from one folder to another inherits all of the policies from the target hierarchy

ETL solutions

	Underlying Technology	Use Case
Dataflow	Apache Beam	Batch and Streaming Data at Scale Log ingestion/transformation AI/ML
Data Fusion	CDAP DataProc	Streaming Data at Scale from various sources into a DWH Regular ingestion into DWH
DataProc	GCE	Spark/Hadoop Cluster Distributed processing of large datasets
Dataprep	Apache Beam	Data visualization, to explore, clean and prep for analysis and ML.

Not being a data scientist and not having enough knowledge on the individual products, use cases or technology makes this one hard to get my head around.