EDGELESS: orchestration domain with 5 VM nodes

Scenario

ε-ORC, ε-CON, Redis in a VM

5 nodes in other VMs:

4 with 2 cores

1 with 8 cores

Factors:

load, in terms of the interarrival between consecutive workflows, with the average ranging from 10 s to 60 s

orchestration policy, i.e., either Random or RoundRobin

Each experiment lasts 1 hour and is repeated 10 times.

Workload generated with edgeless_benchmark, see run.sh for the complete set of parameters.

Repeatability

Check GitHub repository with the EDGELESS reference implementation and experiments (experiment 003-tid-benchmark-varload)

Install a Redis server

Update the ε-ORC and ε-CON configuration in conf/

Install a cluster of EDGELESS nodes

Use run.sh to run experiments by saving the output data in dataset

Dataset content

The compressed archive 003-tid-benchmark-varload.zip contains the following files, all with CSV entries:

Filename
Format

health_status.csv
timestamp,node_id,node_health_status

capabilities.csv
timestamp,node_id,node_capabilities

mapping_to_instance_id.csv
timestamp,logical_id,node_id1,physical_id1,...

performance_samples.csv
metric,identifier,value,timestamp

application_metrics.csv
entity,identifier,value,timestamp

Notes:

The timestamp format is always A.B, where A is the Unix epoch in seconds and B is the fractional part in nanoseconds.

All the identifiers (node_id, logical_id, and physical_id) are UUID.

The field entity in the application metrics can be f (function) or w (workflow).