CS186-L20: NoSQL

Scaling Relational Databases isn’t always the best option

including partitioning and replication

BUT, consistency is hard to enforce!

Taxonomy of NoSQL Data Models

Key-Value Stores

1
2

Map<Key, Value>
get/put

Distribution / Partitioning, just using hash function

• if no replication, key k is stored on $h(k)$ node
• if multi-way replication, key k is stored on $h_i(k), i=1,2,…,n$ nodes

Extensible Record Stores

the idea is that do not use a simple key to lookup 🤔

Document Stores

JSON Documents

using JSON as example

do not store replicated key!

JSON is a Tree 🪵, Self-describing 💬, and Flexible 🔥

can store Json in RDBMS

1
2

SELECT # FROM people
WHERE person @> '{"name": "John Doe", "age": 30}';

mapping between JSON and Relational Data

Relational Data Model ===> JSON Document easy, note that replicated key can be handled by using a array [ ]

JSON Document ===> Relational Data Model

• using NULL to represent missing values
• nested or replicated data? hard to handle! multi-tables may help 🤔

Introduction to MongoDB

基本语法

select and find

1
2

db.collection.find(<predicate>, optional<projection>) 
db.inventory.find({}) // return all documents

Limit and sort

MQL Aggregations and Updates

unwind

update

MongoDB internals