Technologies we use

Non-Relational Databases

What is a non-relational database?

A non-relational database, also known as NoSQL database, offers an alternative model to a traditional relational database, whereby you can store, retrieve, and link data. In a relational database, your data is stored in tables and rows, linked by keys, whereas non relational databases come in many different forms of models, and are typically used for fluctuating data sets and requirements, making them perfect for handling big data and real-time applications.

Big data grows rapidly, changes a lot (and changes quickly), and has a diversified structure. Relational database models were not designed with these requirements in mind, so are inefficient at storing, retrieving, and linking this data. On the other hand, the right type of non-relational database can store and manage data with great flexibility, perform high speed operations and scale easily, making it a clear choice for developers in these sort of applications.

The differences between NoSQL and relational databases really depend on your data set and applications.

Types of non-relational databases

There are four basic different varieties of NoSQL databases:

Key-value stores

in which values are accessed via keys in a hash table,

Document-based stores

where key-value pairs are compressed documents, consisting of data encoded in a form such as JSON, XML, etc.,

Wide-column stores

where data is stored in columns (accessible via a row key), not rows as it is in a traditional SQL system,

Graph stores

where data is represented by a network of graph.

NoSQL databases also can be based on a schema, or schema-less. A schema-less database doesn’t require you to follow a strict way of modelling the data, such as predefined datatypes or having the data stored in a tabular form.

There are plenty of use-cases for implementing schema-less NoSQL databases:

Accessing data quickly and in a simple way

rather than focusing on the reliability of transactions,

Storing large volume of data

without wanting to lock oneself into a schema that could be later difficult to change,

Uploading data from multiple sources

and wanting to store it in its original form for maximum flexibility,

Storing data in hierarchical structures

that are described by data itself, without any external schema.

Non-relational databases can be excellent data stores for modern applications. Do you need NoSQL in your business? Get in touch.

Let's get in touch

The benefits of using NoSQL databases

What benefits can using a NoSQL database bring to a project? First of all, it can improve productivity, when the use-case is properly aligned to the particular NoSQL database being used - instead of trying to fit a square peg into a round hole, which causes development delays. Non-relational databases can help improve performance in data access, especially when dealing with large volumes of data; reducing latency and improving throughout. Speaking of large volumes of data, as we already mentioned, NoSQL databases can be perfect for Big Data systems.

Various types of NoSQL databases can be elastically scalable, easily taking advantage of new nodes while transparently expanding. These are usually designed to do this while taking advantage of using low-cost commodity hardware. NoSQL databases can thus be a low-cost database storage solution, as the cost per gigabyte storage can be much lower than the cost of using a relational database storage solution.

MongoDB

MongoDB is a free, open-source, cross-platform document-store type NoSQL database, which stores JSON-like documents, and is geared towards flexibility. Within MongoDB, objects are mapped to application code, making it easy to work with the data. MongoDB comes with indexing, ad hoc queries and real time aggregation features that are used for accessing and analyzing data. MongoDB at its core is a distributed database, making it a high availability NoSQL database with scalability a key benefits.

Some companies that use MongoDB or have relied on it in the past are: Google, Cisco, eBay and Forbes. In 2011, Craigslist chose to migrate more than two billion documents to MongoDB. The company wanted to achieve the scalability that MongoDB is famous for, as well as have flexibility in the schema. This is because each post (with metadata) is stored as single document and MongoDb can accommodate changes as the schema is updated - with limited changes necessary.

Redis

Redis is another popular open source database solution, one of the alternatives to MongoDB. Redis operates in-memory, and can be used as a cache and message broke. It supports various data structures, among them strings, lists, sets, hashes, and bitmaps. This NoSQL database can run atomic operations, for instance appending to a string or computing a set intersection. Redis’s operation with an in-memory data set makes it possible to achieve outstanding performance. The database also offers data replication, which sets up master-slave cache nodes, with slaves updated asynchronously. Its caching abilities can ensure uninterrupted service. Great performance and high availability makes Redis perfect for handling big volumes of various data in any business. Additionally, Redis supports transactions and optimistic locking, as commands can be executed in a batch, not just one at at time.

Redis is, or has been used for NoSQL databases in commercial applications like Twitter, Snapchat and Craigslist. Verizon Communications recently switched over their social channels database to Redis Enterpise, which resulted in huge cost savings and reduced downtimes. Similar, but alternative databases to Redis include AeroSpike and Hbase.

ElasticSearch

ElasticSearch is a distributed, open source content search, data analytics and query engine developed with Java and built on top of Lucene, a full featured information retrieval library. It is schema-free, relies on JSON documents and indexes all fields by default. ElasticSearch is able to solve a growing number of use cases and provides the best full-text search option on the market. ElasticSearch is very flexible. It works by detecting the underlying data structure and then indexing the data in order to make it searchable. The tool is document-oriented and boasts many features, like customized text splitting text, faceted search, and customized stemming. It is high performance and works in real-time.

ElasticSearch is used (or has been used previously) by companies such as Mozilla, Foursquare, GitHub, Uber, and SoundCloud. Autopilot have built a lightning fast marketing campaign insights tool for their customers with the Elastic Stack (ElasticSearch + Logstash + Kibana) that they say would have been impossible to do without it. South East Asia’a biggest ridesharing platform, Grab, uses ElasticSearch to power their 300 million queries per day among 77 million customers. ElasticSearch alternatives include Solr and Amazon CouldSearch.

Do you think that a scalable, performative, low-cost, high availability noSQL database might be right your project? Are you dealing with Big Data? Contact us and we will show how you can benefit from the use of a NoSQL database and choose the technology most suitable for your application!

Let's get in touch