Understanding Database 10 - Data modelling with MongoDB

Modelling for RDBMS 

1. Define schema

2. develop the application and DB queries

-> One correct solution, no revision/Data duplicates /Usage makes schema demoralised(Performace dropping)

 

Modelling for MongoDB

1. Develop and define the data model

2. re-define data model

3. Improve the application

4. improve the data model 

5. repeat cycle 3 and 4

-> Multiple design options/design for usage patterns/Design evolution easy and intuitive (possible without downtime)

※ Things to consider: Data model is defined at the application level/Design is an integral part of each phase of the application lifetime/affected by changes in the app's data need and read and write data usage. 

 

Data Modelling Method

1. Evaluate Application workload 
2. Map out entities and their relationships
3. Finalize data model for each collection 

Evaluate Application workload	Map out entities and their relationships	Finalize data model for each collection	Outcome
- by a business domain expert - consider the current, predicted and future scenarios - consider production logs and matrix - consider the data size  - consider the operations lists ranked by importance(speed, accuracy, etc)	- Link or Embedded?  (by Collection Relationship Diagram- CRD)	- considering data size - considering database queries and indexes - considering current operations and assumptions - identify and apply relevant design patterns	- Data size  - Collection with fields and shapes  - queries and index - assumptions and future/growth projections
** Operations? ** Cardinality?	** How often the embedded info be accessed? ** embedded info changes often? ** queries used embedded data? ** embedded data to be sued together?	** Schema versioning ** Bucket Pattern ** Computed pattern

 

Design Patterns: Schema versioning 

 Schema can be revised, and then instead of updating the entire fields, you can write Schema_version: 2.0 with newly added fields.

MongoDB document

Design Patterns: Bucket pattern

- reduce index sizes by a large magnitude, increase the speed of retrieval of related data 

- From the below figure, if the "valcount" reaches 200, a new document is created automatically and starts storing in the new document. 

MongoDB Document

Design Pattern: Computed Pattern

- instead of using the computed result in the document, keep the total count and qty(summary records)

- reads are more common than writing, compute on write is less work than compute on read, kinds of caching pattern

MongoDB document

-> MongoDB Youtube provide an excellent explanation for this part.