Design Part 2

Short URL

In Twitter, Weibo, you have charactor limit. Also normal URL will contain some invalid ASCII code that you cannot put into the content.

So a lot of company has its own Short URL

Example: http://t.cn/RGKGR3T

Short URL

Design A System that can generate this Short URL

The user input an URL, the server returns a short one
Other user click this short URL, it will redirect to the real URL
If the URL has not been used (no one clicks for 72 hours, configurable), it will be removed

How to generate a short URL

UUID

MD5

Are these too long? Could we find a better way?

Think about the requirement

one URL could be used released after some day

How to generate a short URL

We do not generate a short URL everytime, we just lookup the table

How to send request

Restful service

Restful API

Design a Database

Short URL -> Long URL, time

Pretty Easy? If we have a high QPS?

What DB we should use?

Relational Database: Oracle, MySql

Non-Relational Database (NO Sql): MangoDB, CouchDB, Couchbase, Redis

RDBMS vs NoSQL

RDBMS(Relational Database Management System)

Strong mathematical basis
Declarative syntax
A well-known language in Structured Query Language (SQL)
Real-time Read
Read/Write Consistency

RDBMS vs NoSQL

NoSQL (Not Only SQL)

It does not have schema
Read/Write throughput is very high
Horizontal scalability can be achieved easily
Will support Bigdata in volumes of Terra Bytes & Peta Bytes
Faster development life cycles for developers
Can be hosted in cheaper hardware machines
In memory caching option is available to increase the performance of queries

Lots of companies combine these two together

For this question

WE could just use NoSQL

Reason: we do not have complicated query to fetch data, we only need original URL and short URL (timestamp maybe)

How to remove unused links?

How to update the Store?

How to remove them?

How to remove unused links?

short -> Long

short -> time

time -> short

Is it okay?

How to remove unused links?

If we could stretch the time

short -> Long

short -> time

Daily/hourly data purge job

RDBMS + NOSQL

Relational Database cannot handle high reading QPS

NOSQL has a lot limit in relation and maintenance

So sometimes we just hybrid them together

RDBMS + NOSQL

For common get queries, we store them in NOSQL

For specific queries and write operation, we still use RDBMS

How to update NOSQL? Database Listener

Summary

Database: Relational and NoSql -> consider to use which one

Network: Apache, Tomcat, etc -> Rest API, JSON

Distributed System: Load balancing, Sticky Session

Security: HTTPS, Encryption/Decryption