How will you know whether your problem is a graph problem?
At one end of the spectrum, enterprises are identifying potential use cases for graph databases. Typically, the use cases revolve around the need for the traversal of a highly connected network and hierarchical datasets in ways that are cumbersome with RDBMSs and NoSQL solutions. At the other end of the spectrum, enterprises are not sure of how and when to leverage graph databases. In fact, they may be asking themselves, “Is my problem a graph problem?” Ted Wilmes, data architect at Expero, while exploring the graph database landscape with a particular focus on JanusGraph, answers this critical question.
To begin with, enterprises should gain an understanding of the two graph-modeling paradigms— property graphs and resource description framework (RDF) triplestores. Property graphs consist of vertices and edges, where the vertices are connected to each other by the edges. These vertices and edges are labeled and can have properties. The RDF triplestores provide a modeling paradigm that is based on triples — subject-object-predicate. Hence, the modeling approach of RDF triplestores and property graphs differ.
Consequently, the next step would be to determine whether an enterprise is using networked data for instances such as monitoring infrastructure, social networks, finance, retail and others. In addition, an enterprise should have use cases that require “graphy” queries. For example, maintaining an accurate picture of the network of “things” being monitored in IoT or when trying to derive the value of a financial instrument having a complex structure of assets nested within assets.
These are the type of use cases where a graph database, will provide greater value over other expressive modeling-wise options, such as a RDBMS. In addition, enterprises would not gain any appreciable benefit in using graphs to resolve everyday SQL queries.
Click here to read the complete article.