The Battle Between Graph Databases: ArangoDB vs Neo4J

The database is one of the must-have components for each software project. In fact, the number of cases, when developers don't need a database is very limited and can be counted on the fingers of one hand. Thus, it is possible to ignore them. Instead, let's consider in detail one of the most recent database battles: ArangoDB vs Neo4j, and try to figure out what are the best circumstances to choose each.

What Do We Know About Databases?

There are two classic database types: SQL and NoSQL. There is a lot to say and explain about the difference between them, however, everything we need to know by now - is that both ArangoDB and Neo4j belong to the NoSQL database type.

First of all, it is impossible not to notice the overall growth of attention to NoSQL databases, especially graph ones. Usually, they are much more flexible and propose multiple out-of-box solutions and interesting tools. Graph Databases are also considered to be more flexible and adjustable than others. However, the clearest and most obvious advantage of this type of DB is the low entry threshold, i.e. most of them are very easy to work with and require less development background. Moreover, on the code level, they are less scale and can help to significantly decrease the number of code lines and the overall code units.

Nevertheless, graph databases, as well as most other NoSQL DBs are used as additional data storage and management tool. Clearly, they have a lot of various advantages, which are related to different fields. Yet, the SQL type of database has a great advantage - it can help to structurize and easily manage already structured data. In other words, traditional relational databases are better for operating relational, predefined, and specified data. NoSQL databases can have countless advantages, however, most of them won't be able to outdo this crucial aspect.

On the other hand, do they really need to? Frankly speaking, most of our database specialists believe that there is no such need. On the contrary, they highly recommend others combine SQL and NoSQL solutions whenever it is useful or needed. Apparently, if such a combination is unnecessary and brings no significant improvements, then there is no need to complicate your work one more time. But some of our cases have proven that sometimes such a mix can become a very reasonable choice.

Summing up, we can conclude that it is mostly better to use the NoSQL type as an additional tool for data management when the relational database is used as a core database. Despite all the advantages and key features, it may be hard to use the NoSQL database model as a source database.

Neo4j Database In a Nutshell

Neo4j is one of the most popular graph database examples with more than 140 million downloads. It is a simple, yet powerful tool which is mainly used to define relations between objects, known also as nodes. Its graph nature simplifies this process, including sub-relations, specifying the relation "by similarities", or "property graph model".

Among easily recognizable we can highlight the graph display. As a matter of fact, this database stores data as the developer entered it. Nonetheless, it displays it in a way of the graph, where it is easy to distinguish the entered nodes, as well as their connections.

As a result, it is not only easy to navigate without any additional actions but also easy to use as any other mature and classic database. Among other helpful features, we can highlight also detailed but easy-to-understand documentation, user-friendly design patterns, and compacted code units, that are much smaller than competitors' proposals.

Briefly about ArangoDB

While most databases are providing a single option, for instance, Neo4j is a graph database type only, ArangoDB is a swiss multitool in this field. It was designed specifically to maintain various types of NoSQL databases. Therefore, ArangoDB use cases may include the development of key-value, graph, or document-oriented databases at the same time. Undoubtedly, it is a great feature, which helps to immediately stand out from the rest of the rivals. However, to embody such an opportunity is only halfway there. It is also important to make sure that all of these variable behavior types would work as supposed to. Looking ahead, we can state that ArangoDB developers managed not only to develop a completely out-of-box solution but to ensure it works fine.

It is a unique native multi-model database technology, that allows developing an app, without previously mentioned, yet a very common approach of combining SQL and NoSQL databases within the project. Anyway, it is worth admitting that despite such a rich variety of other supported types, the main focus of ArangoDB is graph database features. Thus, it is rather a graph DB with extended functionality and improved compatibility with other patterns.

Comparing ArangoDB vs Neo4J

We propose to start with the most obvious differences or similarities, moving to more ambiguous ones. Frankly speaking, these two DBs are commonly compared. However, there is still no obvious winner and each reviewer gives his own humble opinion on the subject.

Functional Capabilities

Taking into account all the above, we can assume that we have a clear winner, at least talking about the variety of possible options. Neo4j provides services, related to graph database type only. At the same time, ArangoDB proposes other NoSQL solutions as a bonus feature. Thus, developers can sometimes use other data storage and management approaches, behaving like document-oriented or key-value databases if needed.

Nonetheless, this statement may seem controversial at first, but both databases are “graph-first”. In other words, the main focus for both is to work according to graph principles. Also, it appears that ArangoDB owners managed to ensure the high quality of their product, so it can confidently compete for the title of the best. On the other hand, Neo4j is more recognizable and is still in great demand. It is a relatively simple yet powerful database, that is capable of most required features, assuring that everything will work as great as possible.

Social Networking

Neo4j database has an obvious 5-year head start. Moreover, it is much more popular and widespread. As a result, there are more potential users, who can take part in discussions or share their experiences. On the one hand, the official website proposes a variety of information apart from the documentation, like "how to" guides and tutorials. In fact, there are even separate sections "Developer" and "Learn", where everyone can find useful info or other insights. Moreover, the Neo4j team regularly posts content on their blog. The number of related content is countless, it includes videos, books, specified libraries, case studies, academies, etc.

On the other hand, there are no forums or conversation units. Thus, the only way to get your content posted - is to get approval from the website owners. As a result, to take an active part in discussions, customers have to look for third-party platforms with corresponding topics.

ArangoDB website proposes mostly identical features. For instance, there are also case and use studies, ArangoDB university, blogs, and so on. Nonetheless, the Arango team also constantly posts comparisons of their product with the competitors. Additionally, there is an opportunity to join an official community, which is divided into various activities and social networks.

Query Language

Each database is using own query language. ArangoDB is operated with AQL, which stands for ArangoDB Query Language. Due to the multi-model nature, the need for a specific query language appeared. While most DBs are operating a single approach, whether it is SQL or NoSQL type, Arango supports multiple types. Thus, to simplify and enable stable connection and allow comfortable operating, developers created AQL, a unified language that supports various data formats or patterns.

It is also a declarative language and is a unique solution in many ways. For instance, despite the fact, that ArangoDB is a NoSQL database type, the language, it uses, has more in common with SQL ones. Franky speaking, the same can be told about Cypher, a declarative graph query language, used in the Neo4j database.

The main difference, though, lies in the final result. For example, Cypher was developed by the image of SQL, yet its creators paid more attention to the requirements of graph databases. As a result, they managed to develop a query language that can comprehend not only data units(nodes) but their relations as well. Additionally, it become one of the most important graph query languages thanks to the OpenCypher project.

Summing up all the foregoing, we can also say that Cypher is not only more crucial for the industry, because nowadays it is used in other products, that are not related to the Neo4j at all. Thus, it means that this language will keep constantly improving by different developers, bringing on different modifications or upgrades. At the same time, AQL is used by ArangoDB only, therefore it may take more time and resources to maintain this query language, as well as the number of variations or upgrades will be more limited, compared with Cypher.

Extensibility Possibilities

ArangoDB supports numerous extension options. Thanks to the Foxx microservices framework, it is possible to create not only synchronous microservices architecture but to develop their own services on demand. Consequently, developers are able to extend the functionality with the use of self-made APIs. Foxx framework allows implementing REST endpoint or even full-fledged applications, that will work within the database. It is also possible to migrate from other JavaScript environments like Node.js.

However, despite such incredible abilities, there are also numerous limitations. For example, async-based processes are incompatible with this framework, so most of them probably must be rewritten. Also, Foxx uses V8 JavaScript Engine, which has its own native boundaries like a 512MB memory limit.

Neo4j's extension process is based on plugin implementations. Thus, developers are able to expand the functionality either by developing their own or by integrating ready-to-use plugins. Among the advantages of such an approach, we can highlight the user-friendly and intuitive procedure of adding plugins. Combined with the numerous community, this extension option can make a real difference and help developers easily customize their projects.

Yet, it is hard to compare the plugin's system with the APIs, because it is probably impossible to figure out how advanced each can be and if there is any functionality gap between these methods. All we are saying - is that both databases can be adjusted and improved with the use of a wide variety of premade solutions.

Summary

We have to admit, that it may seem that such a comparison may seem incomplete. For instance, we skipped some very important benchmarks like performance or data studies, the technical side of both databases, their limits, etc. Honestly, we understand and want to clarify it. Comparing such technologies is not an easy task and it always depends on the circumstances and context of the use cases. The comparison of general info is not an actual option, because it won’t help to make any assumption except for the fact, that these opponents have more similarities than differences. Clearly, it is a wrong assumption, because the practical and technical aspects of both vitals are completely different, yet it is possible to notice them in a concrete example.

On the one hand, each of examined competitors proposes a case studies list, where they explain the usage of their database technologies in different products. Yet, most of them include general info only, without any detailed reports or infographics. Talking about third parties, there are numerous comparison boards and each proposes unique results. Thus, it is impossible to make a final independent and accurate estimation, based on these reports. The one and only way to choose the correct technology - is to discuss your project with the development experts, who are willing to take part in its development.

Remember - there is no bad technology, but there are wrong use cases.