GCP Classroom Series – 07/Apr/2021

NoSQL Databases

• We have four different categories of NoSQL Families
  • Document Databases:
    • stores the data elements in the document like structures (json)
    • Commonly used in content management, monitoring. Web and Mobile applications
    • Examples: Couchbase Server, CouchDb, MarkLogic, Mongo DB
  • Graph Databases:
    • Emphasise conection b/w data elements, storing related nodes in Graphs to accelarate querying
    • Used in recommendation engines and geospatial applications
    • Examples: IBM Graph, Neo4j, Allegrograph
  • Key-value databases
    • Uses a simple data model that pairs unique key and its associated values in storing data elements
    • Examples: DynamoDB, Aerospike, Riak
  • Wide column stores: They are also called as table-style databases
    • store data accross tables that can have very large number of columns
    • Commonly used in internet search and large scale web apps
    • Examples: HBase, Cassandra, SimpleDB
• The NoSQL offerings on the GCP are Bigtable and Datastore

Cloud Bigtable

• This is alternative to the Hadoop’s HBase
• Google describes Bigtable as sparsely populated table that can scale to billions of rows and thousands of colums
• When not to use Big table
  • Do not use bigtable if you require transaction support
  • Do not use big table if your data size is less than 1 TB
  • Do not use big table if you plan to use Analytics, BI or datawarehousing use cases. (BigQuery is a better option here)
  • Do not use Bigtable for very highly structure or hierarchical data
• When to use Big table
  • The first and obvious one has to do with very fast scanning with low latency and high througput applications
  • If you have write operations taht are very infrequent or not import and you dont care about ACID support

Google cloud Datastore

• This is fully managed NoSQL database service.
• This is something that we turn when we are looking for document-oriented storage in NoSQL database
• To use data store from
  • Java Refer Here
  • Python Refer Here

BigQuery

• Data warehouse as a Service offered by GCP is BigQuery.
• BigQuery is Google’s fully managed petabyte scale serverless database
• To play with Bigquery google offers public datasets Refer Here

Memory store

• GCP offers
  • Memory Store for Redis
  • Memory store for Memcached
• Using redis from python Refer Here
• Using redis from Java Refer Here
• We can create a Memory store for Redis and Memcached and use it from applications hosted in Google APP Engine or VM Instances using private ip