Unrivalled Performance

Architecture for this Decade

Existing DBMS products were engineered in a time when processing, memory, storage were expensive and more-limited in their capability. GraphBase. on the other hand, has been built for modern multiprocessor servers and is designed to take maximum advantage of big RAM and high-speed storage.

A single low-cost 1RU GraphBase server can handle billions of queries and updates per day against a graph of one billion vertices and 100 billion arcs. That's FaceBook on a single pizza box.

The secret to this performance is sophisticated thread management, and compact structures that allow as much of the graph as possible to remain in memory. GraphBase also carries a host of unique innovations. Arc heuristics, for example, enable graph traversals and query speed 10 to 100 times faster than other Graph Database implimentations.

Designed for Cloud

Anyone who's worked with graph structures, knows that partitioning or "sharding" graph data is a difficult problem. But sometimes it's not possible to keep your entire graph on one server. At other times it makes sense to distribute your graph so that processing can also be distributed.

GraphBase is designed to be distributed in true "Cloud" fashion. Each server "Node" is autonomous, but aware of it's obligations to it's peers. Communication between nodes is asynchronous, and sophisticated cacheing and queueing strategies allow a group of GraphBase Nodes to accommodate the latency and bandwith issues of a geographically-distributed cloud.

A true Graph Database keeps its arcs close - a strategy refered-to as index-free adjacency. Unlike some competing products, GraphBase arcs are encapsulated within each vertex and go wherever that vertex goes. It's an architecture that greatly simplifies data distribution.

The Only Graph DBMS for Big Data

Graphs are great for simplifying and managing complex data structures, but they're the wrong tool for handling the high-volume kludge of classic "big data" problems.

GraphBase lets you embed simple, compressed, highly-efficent, vertex-focused data stores. Think "all the phonecalls or transactions for a person". It's strategy so effective that it permits a level of real-time big data analysis that's difficult - and expensive - to achieve with any other technology.


Comments

speed. redundancy. scale.