Method for representing a distributed computing system by graph embedding

1. A method of representing a distributed computing system, the distributed computing system comprising a plurality of processing devices connected together in a predefined topology, the method comprising:

receiving at least one piece of data from an activity log file relating to at least one processing device among the plurality of processing devices,

receiving at least one metric related to said at least one processing device among the plurality of processing devices,

receiving at least the predefined topology of the distributed computing system,

constructing a graph representative of a distributed computing system operation, the graph comprising the at least one piece of data extracted from the activity log file that is received, the at least one metric that is received, and the predefined topology that is received,

embedding at least one part of the graph to obtain at least one state vector representing the at least one part of the graph that is embedded, the at least one state vector taking into account the predefined topology, the at least one metric and the activity log file.

2. The method of representing according to claim 1 , wherein the receiving the at least one piece of data, the receiving the at least one metric and the receiving the predefined topology are implemented a plurality of times, each implementation being performed during a time window of predetermined duration, and wherein the constructing the graph and the embedding the at least one part of the graph are carried out at an end of each time window.

3. The method of representing according to claim 1 , wherein the graph that is constructed comprises

nodes representing one or more of users, processing devices, activities and log files,

edges connecting said nodes, where the edges represent links between said nodes and are based at least in part on the predefined topology,

attributes of one or more of said nodes and said edges representing the at least one metric that is received.

4. The method of representing according to claim 1 , wherein the predefined topology that is received is sent by an orchestrator.

5. The method of representing according to claim 1 , wherein the embedding of the at least a portion of the graph is performed by a graph embedding method selected from DyGCN and MAGNN.

6. The method of representing according to claim 1 , further comprising maintaining the distributed computing system by a machine learning method from the at least one state vector.

7. The method of representing according to claim 6 , wherein said maintaining the distributed computing system is carried out from a plurality of state vectors respectively obtained during each time window of a plurality of time windows.

8. The method of representing according to claim 6 , wherein the machine learning method is at least one method selected from

an anomaly detection method,

an incident prediction method,

a root cause analysis method.

9. A device for representing a distributed computing system, the distributed computing system comprising a plurality of processing devices connected together in a predefined topology, the device comprises:

at least one receiver, wherein the at least one receiver is configured to

receive at least one piece of data from an activity log file relating to at least one processing device among the plurality of processing devices,

receive at least one metric related to said at least one processing device among the plurality of processing devices,

receive at least the predefined topology of the distributed computing system,

at least one graph creator, wherein the at least one graph creator is configured

construct a graph representative of a distributed computing system operation, the graph comprising the at least one piece of data extracted from the activity log file that is received, the at least one metric that is received, and the predefined topology that is received,

at least one graph embedding network, wherein the at least one graph embedding network is configured to

embed at least one part of the graph to obtain at least one state vector representing the at least one part of the graph that is embedded, the at least one state vector taking into account the predefined topology, the at least one metric and the activity log file.

10. The device according to claim 9 , further comprising a storage configured to store the at least one state vector from the embed said at least one part of the graph, wherein the at least one graph embedding network is further configured to send the at least one state vector to the storage.

11. A non-transitory computer-readable medium comprising instructions that, when software is executed by a computer, cause the computer to implement a method of representing a distributed computing system, the distributed computing system comprising a plurality of processing devices connected together in a predefined topology, the method comprising:

receiving at least one piece of data from an activity log file relating to at least one processing device among the plurality of processing devices,

receiving at least one metric related to said at least one processing device among the plurality of processing devices,

receiving at least the predefined topology of the distributed computing system,

constructing a graph representative of a distributed computing system operation, the graph comprising the at least one piece of data extracted from the activity log file that is received, the at least one metric that is received, and the predefined topology that is received,

embedding at least one part of the graph to obtain at least one state vector representing the at least one part of the graph that is embedded, the at least one state vector taking into account the predefined topology, the at least one metric and the activity log file.

12. The non-transitory computer-readable medium according to claim 11 , further comprising a computer program product.