NebulaGraph Database: How can I speed it up if a task result is saved slowly or data is transferred slowly between tasks?

Question

When I use NebulaGraph and NebulaGraph Explorer for graph algorithm, it takes a long time. How can I speed it up?

I used the workflow feature in NebulaGraph Explorer, which allows me to design my own workflow, but once I did, the execution was slow.

Answer 1

If you are only exploring how you would like to finally implement your production graph-based pipeline/application with graph algorithms, quickly execution and experiments will be extremely helpful(instead of running algos on the whole graph for each epoch of your quick study).

It's doable in explorer in may ways:

1. consider create a smaller dataset for the study
2. use query(with limit sampling of the data) instead of scanning full data in the workflow pipeline
3. explore and query the subgraph of the data in the canvas and leverage the explorer's in-browser graph algorithms, everything will be run on small piece of data visually on your browser

I personally will go with option 3 to do fast back-and-forth graph algo experiments first and then go with workflow before final implementations.

As for option.1 it'll be not easy to cut a smaller piece of graph while with the expected graph structure/connectivity unchanged, plus option.2 is with a similar cost of the option.3, but we could do everything with drag-drop-click in the option.3 :).

ref:

• https://docs.nebula-graph.io/3.3.0/nebula-explorer/graph-explorer/graph-algorithm/ the in-browser graph algorithm function