Sizing and scalability considerations
This topic provides information about performance and scalability test results for PATROL for Kubernetes 1.0.00.
PATROL for Kubernetes can successfully monitor approximately 1100 pods on 12 node cluster using one PATROL Agent.
The following tables list the configuration that was used for running the performance and scalability tests.
Service configuration
Platform | Processors | Memory |
|---|---|---|
Microsoft Windows Server 2012 R2 Datacenter | 4 | 8 GB |
Setup details
KM | PATROL Agent | OpenJDK | TrueSight console |
|---|---|---|---|
PATROL for Kubernetes 1.0 | PATROL Agent 11.3.01 | 11.0.2 | TrueSight Presentation Server 11.3.01 |
Configuration
Max Heap Size | Min Heap Size | JMX remote | javaOpts |
|---|---|---|---|
1024m | 256m | Enabled | -Xms256m -Xmx1024m |
Monitored environment details
Entity | Count |
|---|---|
Pods | 1100 |
Container | 1350 |
Deployment | 545 |
Performance, Scalability, and Reliability
Scalability
Total PATROL Instances | Total PATROL Parameters | PATROL Agent – Private Bytes | Java process – Private Bytes |
|---|---|---|---|
14 K | 54 K | 179 MB | 1.13 GB |
Resource Utilization
Performance data of the PATROL Agent and Java collector
Process | Average CPU in % | Average working set |
|---|---|---|
Java Collector | 1.8 | 1105 MB |
PATROL Agent | 2.5 | 186 MB |
JVM resources
Average Heap Utilization | Max Heap Utilization | Peak Thread Count | Thread Count | Daemon Thread Count |
|---|---|---|---|---|
600 MB | 720 MB | 251 | 23 | 18 |