Fastest way to map nodes and edges into RedisGraph

Redis modules RedisGraph
1

Hello all,
I have hashes structured as node{id}={‘id’:id, ‘name’:name, ‘rank’}
and edges{id}:source_id, destination_id=, sentence_id, rank=int.
They map directly into RedisGraph nodes and edges, what would be the fastest way to batch map from Redis into Redis graph?
I am worried about:
redis_graph.commit() in redisgraph.py example.
Does it mean Graph can only be populated in batches?

1 Like
2

Hi,
In case your data is already in Redis LUA Script might be a valid option reading data from hash and populating a graph key,
Another option would be reading the data out of Redis, constructing a set of CSV files complying with our Bulk-Loader utility.

In case you’ve only have a handful of entities then either RedisGraph client will do the trick,
Lastly a graph is not restricted to batch loading.

1 Like
3

Hello @SWilly22 thank you for reply.
Do you have example of Lua script which does something similar?
I am trying to use python redisgraph client “on-demand” but the number of entities is fairly large - mostly edges, I would rather prepare hash first and bulk load from. I have seen BuilkLoader script and looking at source code right now, but it seems a bit nonoptimal dump redis into CSV only to load back to Redis.

4

I’ll suggest starting here
basically you should be able to use LUA to read hash keys and issue GRAPH.QUERY calls.

1 Like
5

@SWilly22 thank you for your suggestion. One follow up question: do I have to check for the existence of node or edge inside RedisGraph before issuing GRAPH.QUERY CREATE command? In other words - if duplicate crated requests are taken care automatically, do they have performance penalty?

6

To avoid duplicates use the MERGE clause.