Google Cloud Platform - Managed Database Services

Managed Database Services

GOOGLE CLOUD PALTFORM

SQL DB
Cloud SQL is a service for fully-managed and reliable My SQL and PostgreSQL databases.
It is a Regional / Multi-regional / Global service.
This compares to Amazon's relational database service (RDS)
It supports automatic replication, backup, and failover.
If you need to scale a cloud SQL instance you have to do that manually or switch to Cloud spanner

SQL db in parallel
It is the first horizontally scalable, strongly consistent, relational database service.
This is for seriously large systems.
Manually we will end up using Sharded MySQL instances or CockroachDB
It is a Regional / Multi-regional / Global service.
We can use cloud spanner like a normal SQL database without having to give up some of the acid guarantees.
This can scale from one to hundreds or thousands of nodes and Google recommends that you have at least three nodes for any production environment.
Even with a single node the data is very durable with lots of replicas because a node is not actually a single server somewhere.
A node is a server in each of the replication locations.
if you want to be able to read the data with the lowest possible latency, clouds banner offers something called stale reads which are also stronger than the eventual consistency.
Cloud spanner gets away with strong guarantees about consistency per CAP theorem and still keeping partition tolerance is that it gives up 100% availability, but this is still a very very high availability system.
Service level agreement has a 99.999 (five nines) when cloud spanners deployed has a multi region instance.
This system is not based on failover, It is designed so that any of the servers can handle any of the requests.

Column-store SQL
BigQuery is a server less column store data warehouse for analytics using SQL.
It is the most significant service in the Google cloud platform.
This is Regional / Multi-regional service for seriously large systems.
Bigquery do share some use cases with Amazon Redshift (Column Store Datawarehouse)
Bigquery is serverless and offers amazing scale without having to provision anything up front.
It scales internally so it can scan terabytes in seconds and petabytes in minutes.
AWS Athena service in 2016 (Based on Facebook Presto) is effectively copy of BiQquery.
Very high-volume customers can pay a flat rate for BigQuery slots
BigQuery remembers the queries and result that you run and if you wind up making the same query again, against the same data, output will be from its cache for free {query results are generally cached for 24 hours}.

Column store No SQL
It is a low latency and high throughput no SQL database for large operational and analytical applications.
This is a wide columnstore type NO SQL database based on ApacheHbase, and similar to amazon's DynamoDB and Cassandra, If you are willing to use no SQL to scale to really high volumes.
It has also added support for the open source HBase API.
Cloud BigTable also integrates well with Hadoop and also the cloud dataflow and cloud data proc services.
Cloud BigTable can scale without interrupting its functionality and the storage just scales automatically as you add more data.
The cloud BigTable is really made huge workloads and if you don't have that level of scale you might want to consider Cloud Datastore , a managed and auto scaled no SQL database

No SQL
It is a is a managed and auto scaled no SQL database with indexes queries and acid transaction support.
Cloud Datastore would compare to Amazon's DynamoDB service. And if you were doing this on your own you might wind up managing your own MongoDB cluster.
This is a no SQL database so the queries can get a little complicated.

Firebase real time and Cloud fire stores
Virtually real-time client updates
Firebase DB revolves around a single, potentially huge JSON document
Cloud fire stores multi-regional and it has collections which have documents, and those documents contain data.