Image data must be stored in a single data store at minimum cost.
Note: Azure Blob storage is Microsoft's object storage solution for the cloud. Blob storage is optimized for storing massive amounts of unstructured data. Unstructured data is data that does not adhere to a particular data model or definition, such as text or binary data.
Blob storage is designed for:
* Serving images or documents directly to a browser.
* Storing files for distributed access.
* Streaming video and audio.
* Writing to log files.
* Storing data for backup and restore, disaster recovery, and archiving.
* Storing data for analysis by an on-premises or Azure-hosted service.
Reference:
https://docs.microsoft.com/en-us/azure/storage/blobs/storage-blobs-introduction

Explanation
Azure Cosmos DB is Microsoft's globally distributed, multi-model database. Azure Cosmos DB enables you to elastically and independently scale throughput and storage across any number of Azure's geographic regions. It offers throughput, latency, availability, and consistency guarantees with comprehensive service level agreements (SLAs).
You can read data from and write data to Azure Cosmos DB using Databricks.
Note on fluid schema:
If you are managing data whose structures are constantly changing at a high rate, particularly if transactions can come from external sources where it is difficult to enforce conformity across the database, you may want to consider a more schema-agnostic approach using a managed NoSQL database service like Azure Cosmos DB.
Reference:
https://docs.databricks.com/data/data-sources/azure/cosmosdb-connector.html
https://docs.microsoft.com/en-us/azure/cosmos-db/relational-nosql

Explanation

Box 1: Hash-distributed tables
The new Health Insights application must be built on a massively parallel processing (MPP) architecture that will support the high performance of joins on large fact tables.
Hash-distributed tables improve query performance on large fact tables.
Box 2: Round-robin distributed tables
A round-robin distributed table distributes table rows evenly across all distributions. The assignment of rows to distributions is random.
Scenario:
ADatum identifies the following requirements for the Health Insights application:
* The new Health Insights application must be built on a massively parallel processing (MPP) architecture that will support the high performance of joins on large fact tables.
References:
https://docs.microsoft.com/en-us/azure/sql-data-warehouse/sql-data-warehouse-tables-distribute

Explanation

Box 1: DataFrame
DataFrames
Best choice in most situations.
Provides query optimization through Catalyst.
Whole-stage code generation.
Direct memory access.
Low garbage collection (GC) overhead.
Not as developer-friendly as DataSets, as there are no compile-time checks or domain object programming.
Box 2: parquet
The best format for performance is parquet with snappy compression, which is the default in Spark 2.x.
Parquet stores data in columnar format, and is highly optimized in Spark.

Explanation

Box 1: Self-hosted integration runtime
A self-hosted IR is capable of nunning copy activity between a cloud data stores and a data store in private network.
Box 2: Schedule trigger
Schedule every 8 hours
Box 3: Copy activity
Scenario:
* Customer data, including name, contact information, and loyalty number, comes from Salesforce and can be imported into Azure once every eight hours. Row modified dates are not trusted in the source table.
* Product data, including product ID, name, and category, comes from Salesforce and can be imported into Azure once every eight hours. Row modified dates are not trusted in the source table.