Skip to main content

Anti Patterns

Unbounded Arrays

  • an array without a size limit
  • used subset pattern or extended reference pattern

Bloated Docments

  • A document that groups together related data that is accessed separately
  • say a book store has two pages: home and detailed page. Home shows 10 most popular books, each book displays title and author info. Detail page shows more data for the book, description, SKU, reviews, etc.
  • if we use one collection to store the book details, and use that collection for the home page, it's bloated because home page only need title and author.
  • a better design would be to create another collection to store title and author only. and the original collection can be used for the detailed page.
  • yes it creates data duplicates, but again, it's a trade-off between duplicate data and performance.

Massive Number of Collections

  • 5k per M10 cluster
  • 10k per M20/M30 cluster
  • 10k per replica set
  • 10k per shard

solution:

  • update schema design
  • sharding

Unnecessary Indexes

  • hide them or drop them if they're not used or rarely used

Data Normalization

  • separates data that access together into different different collections
  • use subset pattern or extended referenced pattern

Case-Sensitivity

  • MongoDB is case-sensitive by default

use Collation

  • language specific rules
  • determine how characters are compared and sorted
  • specify locale for desired language, e.g. locale: 'en'
  • strength 1-5 (default is 3) (1-2: case insensitive, 3-5: case sensitive)
  • collation can be set to collections

reges with \i to do case insensitive search is NOT recommended because it's not efficient.

Use collation instead:

db.Books.find(
  {title: 'Practical MongoDB Aggregations' }
).collation({locale: 'en', strength: 2})

Sometimes indexes are case sensitive, so search wouldn't use the indexes to search. To fix it, declare the index with collation:

db.Books.dropIndex({ author: 1 });
db.Books.createIndex(
  { author: 1 },
  {
    collation: {
      locale: 'en',
      strength: 2
    }
  }
);
Back to top