Previous approach that didn't work: Run Stress-NG as CNF, and load stressng as much as possible.
| Category | Details |
|---|
| Cloud* | Kubernetes-cluster |
| Workload | Nginx, vRouter, ... |
| Traffic-Generators | iperf, netperf, wrk2, trex/prox (sriov) |
| Impairment-Tool | Litmus |
Types of Impairments
Steven Casey : Run experiments one at a time | POD_Level Container Kill
Disk Fill
Pod Autoscaler
Pod CPU Hog Exec
Pod CPU Hog
Pod Delete
Pod Dns Error
Pod Dns Spoof
Pod IO Stress
Pod Memory Hog Exec
Pod Memory Hog
Pod Network Corruption
Pod Network Duplication
Pod Network Latency
Pod Network Loss
Pod Network Partition | NODE_Level Docker Service Kill
Kubelet Service Kill
Node CPU Hog
Node Drain
Node IO Stress
Node Memory Hog
Node Restart
Node Taint |
| Metrics | - Infrastructure
- TGen (analyze the impact of Impairments), Litmus-data.
- CNF - External (kubelet)
- CNF - Internal (this is not feasible in many cases) *** If CN-CNF it is supposed expose metrics (/metrics) (micro-services approach)
|
| Metrics-Collection | Prometheus |
| Logs/Events | Kubernetes System |
| Logs-Collection | Elasticsearch |
| Pod | Intel Pod18 (3-Master Nodes and 2 Worker Nodes) |
| Duration | 14 Feb - 28th Feb 2022 (It may get delayed to Intel VPN Issues.) |