Proposal about GraphScope monitoring

To investigate and propose the following content for monitoring

1. Metrics need to be collected
1. CPU, Memory, Disk, Network
2. Number of Sessions, duration of each session.
3. Detailed metrics about Internal data structures, like vertex map size, fragment size, etc.
4. Graph related, like number of vertices, edges, the schema, etc.
5. Query related, like running time of algorithms, loading time, export result time, running time of gremlin query,
6. Status summary, number of tasks, number of running / success / failing tasks
7. Log persistence

2. Management
1. GraphScope Dashboard
2. Management of queries, e.g. watch status, CPU / memory usage, queueing and canceling
3. GraphScope Server as a service for visualizing and managing multiple GraphScope clusterings.

3. Monitoring tools
1. Prometheus

4. Others
1. Surveying what the customers cares about continuously, from a user's aspect and from a SRE's aspect
2. Take care of both GraphScope and GraphScope-Store