GraphBase Q & A

Please ask your questions here. We'll address each question and add it to this page.

Q. What are the absolute limits on number of Nodes and Relationships?
In theory, none. Testing tells us that a simply-structured graph of a billion vertices and 100 billion arcs can be managed on a high-end 1RU server. The degree to which larger graphs can be distributed over multiple servers will be determined by the nature of your graph.

Q. Can all properties of a vertex (node) or arc (relationship) be indexed?
This depends on the graph structure chosen. For Graph Simple Form, the answer is Yes. GraphBase supports two quite different indexing strategies; one is "global" to a super-graph and the other can be applied by a query during traversal.

Q. Can we preload vertices (nodes) or arcs (relationships) from existing sources?
Yes. Check out our RapidGrapher.

Q. How does GraphBase store data?
We use our own serialization framework for storage, replication and data distribution. Data appears on-disk as large “slab” files with paired indexes.

Q. Can GraphBase be sharded to support horizontal scaling?
Yes it can. GraphBase is designed to be distributed across multiple "Nodes" - and those Nodes can be geographically-separated. But you should always try to avoid sharding if you can - "cutting up" a highly-connected graph is never a trivial exercise. GraphBase uses compact in-memory structures to significantly reduce your need to distribute a graph.

Q. How well does GraphBase scale vertically? Can it make use of "heavy iron"?
The GraphBase architecture reflects this decade's hardware capabilities. It was designed to make best-possible use of multiple processors and large amounts of shared memory.

Q. We're a struggling technology startup, can you help us out?
Yes we can. Tell us about yourselves and what you're planning using our contact form and we'll get back to you.


Comments

THE next generation database