Google Cloud Platform (GCP) Cloud Datastore Product Overview
Google Cloud Platform (GCP)
Features
Global namespace for buckets
Buckets are regional or multi-regional
Bucket Properties
Durability
Availability
Classes - frequency of anticipated access
Long object names - forward slashes mimic file system structure
Do NOT put sensitive data in the name
Path segments are NOT folders - so renames need to be done to all objects individually
Objects are NOT files - they are blobs - no support for "append"
TLS encryption in transit
Distributed service - no single point of failure
Higher latency than HDFS on a Persistent Disk in the cluster
Higher throughput than HDFS on a Persistent Disk in the cluster
Tips
Avoid small reads
Use large block sizes
Avoid iterating over many nested directories in a single job
Access Pattern
Summary: Use as a transactional, No-SQL, object-oriented database
Capacity: Terabytes
Access metaphor: Persistent Hashmap
Read: Filter objects on a property
Write: PUT object
Update granularity: Attribute
Usage: Structured data from AppEngine apps