A Social blog using MongoDB ITEC-810 Final Presentation Lucero Soria - 42403871 Supervisor: Dr. Jian Yang Agenda • Introduction • Methodology • Outcomes • Blog implementation • MongoDB vs. Relational databases • Conclusions 2 Agenda • Introduction • Methodology • Outcomes • Blog implementation • MongoDB vs. Relational databases • Conclusions 3 Problem Specification Relational Databases Management Systems (RDBMS), such as MySQL, do not provide the flexibility and scalability needed to manage social media data NoSQL databases, such as MongoDB, emerged to provide the features that modern applications demand such as flexibility, scalability and productivity 4 Project Aim Analyse the differences between MongoDB and relational databases, especially in supporting social media data 5 Background Sources MongoDB • MongoDB Online Manual • Online articles Relational databases • MySQL 5.5 reference manual • Social Media Management Handbook by Robert Wollan • Online articles 6 Agenda • Introduction • Methodology • Outcomes • Blog implementation • MongoDB vs. Relational databases • Conclusions 7 Project Approach This project is a combination of analysis and development tasks Research MongoDB, social media data and relational databases Implement a social blog using MongoDB Based on the implementation and research: Analyse the differences between MongoDB and relational databases 8 Methodology Incremental methodology was used to implement the social blog • Combines waterfall model with iterations 9 Agenda • Introduction • Methodology • Outcomes • Blog implementation • MongoDB vs. Relational databases • Conclusions 10 A social blog with MongoDB Features implemented: •Login with facebook to create user’s profile in MongoDB •Create, edit and delete posts (text, photos or videos) •Add comments •Search by tags •Sort by blogs with more comments 11 Analysis Based on our experience implementing the social blog, the most relevant features to manage social media data are: •Handle irregular data •Handle large binary objects (videos, photos) • Operations • Metadata •Manage huge volume of data •Handle geospatial queries 12 Relational data model • Fixed-schema • Assume well-defined structure data with a fixed number of fields (columns) and relationships • Minimize redundancy and dependency Normalization 13 Source: http://blog.jruby.org/ Terminology RDBMS MongoDB Table Collection Rows JSON Document Index Index Join Embedding & Linking 14 Document-oriented data model MongoDB uses a document-oriented model using collections Main characteristics: • Schema-less • Collections can be created on-the-fly when first referenced • Capped collections: Fixed size, older records dropped after limit reached • Collections store documents 15 MongoDB Document Main characteristics: • Are represented in a format called BSON (Binary JSON) • Data is de-normalized • No joins Embedding & Linking { author: ‘Lucero', created: Date(‘06-06-2012'), title: 'Yet another blog post', text: 'Here is the text...', tags: [ 'example', ‘lucero' ], comments: [ { author: 'jim', comment: 'I disagree' }, { author: 'nancy', comment: 'Good post' }] } 16 Storing irregular data Example: Different information in user profiles MongoDB • Each document can have different information doc1 = {name: “Joe”, age: ”20”, interest: ”football” } doc2 = {name : “Michele”} Relational database • Tables with all attributes • NULL value in columns where data was not provided Results: Special queries to handle NULL values Expensive 17 Managing large binary data MongoDB • • • • Divide a large file among multiples documents (GridFS) Include metadata to large files Search files base on its content Retrieve only the first N bytes of a video Relational database • Use BLOB (Binary large objects) • Inefficient manipulating rich media • BLOB cannot be searched or manipulated using standard database command 18 Geospatial Indexes Queries to find the nearest N point to a current location MongoDB • Embedded Geospatial features Relational database • Spatial extensions • MySQL implements a subset of the SQL with Geometry Types environment proposed by Open Geospatial Consortium (OGC) 19 Managing huge volume of data MongoDB • High performance • No joins and embedding makes reads and writes fast • Indexes including indexing of keys from embedded documents and arrays • Horizontal scalability • Automatic sharding (auto-partitioning of data across servers) Relational database • Have shown poor performance on certain data-intensive applications and delivering streaming media Case study: Foursquare • Difficult to scale to multiple servers 20 Agenda • Introduction • Methodology • Outcomes • Blog implementation • MongoDB vs. Relational databases • Conclusions 21 Conclusions Benefits that MongoDB offers over relational database: • • • • Flexible schema High performance Manipulation of large object files out of the box Embedded geospatial features However, • MongoDB does not replace relational databases • MongoDB and relational databases can coexist 22 Thank You! Q&A 23