Trinity – the New Microsoft Database Technology

Question, how do you access data from a large database collection, where there are no easy facts to search from? How many sales were in the 200,000 range or higher category? That question can be easily answered from a database because the collection of data is based on facts. But what if there are no easy facts to build your query? That’s where Trinity comes in.

The Trinity Graph Database

But Trinity is a graph database and graph computation platform. Central to Trinity is a distributed RAM-based key-value store. An example of a graph database is a social media system, like Linked in or Facebook. As a social graph it has over 800 million nodes and 104 billion edges. The World Wide Web contains over 50 billion web pages and one trillion unique links. But try to get facts from that collection of nodes is not easy, and it can be time consuming. So as an all-in-memory key-value process, Trinity provides fast random data access. This feature naturally makes Trinity suitable for large graph processing.

 

 


Trinity is a graph database from the perspective of data management. It is a parallel graph computation platform from the perspective of graph analytics. And as a database, it provides features such as data indexing, concurrent query processing, and concurrency control. As a computation platform, it provides vertex-based parallel graph computation on large scale graphs.

 

Data access on graphs has no locality meaning that the database system is not structured, like a SQL database. This means that graph exploration incurs massive random data access. And it makes large graph processing very challenging. To address this issue, Trinity makes graphs resident in a distributed memory storage. Then it optimizes the use of main memory and communication to deliver the best performance for both on-line query processing and off-line graph analytics.

The architecture looks like this.

What this portends for data analysis is the use of a new type of database system that has a large boundaries, not those formally created, but randomly created throughout the Internet, or on local applications.

Source The Insider, Microsoft Research

Subscribe & Connect

Share This Post: 

Subscribe to our e-mail newsletter for updates:

No comments yet.

Leave a Reply