08 No SQL

Question 1

benefits of relational databases

Answer 1

1. consistency, concurrency, recovery
2. mature eco system
3. centralised
4. fast
5. easy management

Question 2

limitations of relational databases

Answer 2

1. poor scalability
2. not very robust
3. not distributed (data is geo-distributed)

Question 3

what is no sql

Answer 3

1. non relational db

2. simplified data models to improve the generalisability

Question 4

types of no sql

Answer 4

1. key-value
2. document
   3 columnar
3. graph

Question 5

no sql: key-value

Answer 5

1. simplest
2. key value stored in a hash table
3. efficient, no complex query filters

Question 6

no sql: document

Answer 6

1. data stored in json format

2. 1 json is one document

Question 7

SQL - no SQL migration schemes

Answer 7

Do we really want to migrate?

1. one to one mapping
   - most popular but least effective
   - every table is a collection
   - each tuple is a document
2. denormalisation
   - consolidate small tables to one big table
3. forget migration, design from beginning
   - suitable for big application changes

Question 8

no sql: columnar

Answer 8

1. collection of key value pairs
2. row key, column key, timestamp
3. data compression

Question 9

no sql: graph

Answer 9

1. explicit graph structure
2. each node knows the adjacent nodes
3. apply graph based algorithms easily

eg. 
node
relationship (belongs to)
property ('alice')
label (person)

Question 10

when to use no sql

Answer 10

1. rdbms too restrictive
2. no need for ACID
3. data too large
4. diverse data sources
5. high requirements for data availability