Does ArangoDB have strict maximum number of nodes and streaming API

[Artur Bańkowski]

Both OrientDB and Neo4j have strict limits of maximum number of nodes and relations that can be stored in a graph. How is that in ArangoDB? Unfortunately cannot find any explicit information about.

Does ArangoDB have a streaming api (in Java) like OrientDB, Neo4j and may nother database engines? I do have read queries that return millions of nodes and streaming the result is the only efficient solution.

[Manu]

2) afaik no, see
https://github.com/arangodb/arangodb/pull/1612#issuecomment-199252932
https://github.com/arangodb/arangodb/issues/602 (Add support for websockets)

u could build  a kinda of streaming with a cursor, its then more like batched streaming.

[Wilfried Gösgens]

Hi Arthur,
ArangoDB doesn't limit the reply or graph size as OrientDB and Neo4j do.

Return cursors are built in main memory at query time, you then can pull chunks from that cursor in subsequent HTTP requests.

Usually you would use i.e. the pattern matching traversals to reduce the size of the result set.

So the limit is is determined by the amount of memory your Environment can provide to buffer the cursor on the database size; once that size is exceeded performance will drop.

You then can pull handy chunks from that cursor to work with in your application.

Cheers,
Willi

[Artur Bańkowski]

Hi Wilfried,

Thank you for very exhaustive reply. Do you know about any equation or ready to use calculator to estimate the memory usage?

Cheers,
Artur

Wilfried Gösgens wrote:

--
You received this message because you are subscribed to a topic in the
Google Groups "ArangoDB" group.
To unsubscribe from this topic, visit
https://groups.google.com/d/topic/arangodb/qtDtqCWigFk/unsubscribe.
To unsubscribe from this group and all its topics, send an email to
arangodb+u...@googlegroups.com
<mailto:arangodb+u...@googlegroups.com>.
For more options, visit https://groups.google.com/d/optout.

[Wilfried Gösgens]

Hi Artur,
sounds like an interesting Idea; what exactly do you want to estimate?
 - the memory usage of a result cursor?
 - the memory usage of collections and their indices?

Cheers,
Willi

[Artur Bańkowski]

Hi Willi,

I would like to estimate both:

Collection memory usage with indices, for a graph with billion vertices and seven billions of relations (no properties for a start).

Cursor with two millions of vertices from traversal.

Cheers,

Artur

[Simran Brucherseifer]

In regards to the maximum number of relations, you might be interested in reading this blog post about super nodes and ArangoDB's hybrid index:

https://www.arangodb.com/2016/04/index-free-adjacency-hybrid-indexes-graph-databases/