pxGrid Cloud Performance and Scale Characterization
Scale and Load Requirements for a Single Tenant
- Number of partner applications per tenant = 4
- Number of endpoints per tenant = 400,000
- Onboarding rate for endpoints = 300 transactions per second (TPS)
ISE Deployment Models tested for Performance and Scale:
- Secure Network Server 3695 & Secure Network Server 3615: Five Node Cluster
- Secure Network Server 3655: Two Node Cluster
- Secure Network Server 3615: Two Node Cluster
- Secure Network Server 3615: Single Node Cluster
Performance and Scale results
Messaging or event load generated due to onboarding of enpdoints
| S.No | Endpoints | Onboarding Rate (in TPS) | Latency (in seconds) | Description |
|---|---|---|---|---|
| 1 | 100,000 | 300 | 510 | Latency is the total time required for the endpoint events to reach the partner applications (including the transit or processing time and the message wait time at each hop). For example, the latency for events generated by 100K endpoints is 510 seconds. |
| 2 | 200,000 | 300 | 865 | |
| 3 | 300,000 | 300 | 1142 | |
| 4 | 400,000 | 300 | 1400 |
API load measured for onboarding 400K endpoints
| S.No | APIs | RPS (requests per second) | Latency (in seconds) | Description |
|---|---|---|---|---|
| 1 | Mix of Open APIs, ERS APIs and pxGrid APIs | 4 | 1 to 8 | The API request load consisted of a combination of ERS-APIs, pxGrid APIs and Open APIs. The API request load was sent by four partner applications in equal ratio. No additional processing or transit delay was observed within the ISE components for this load. For different ISE APIs, based on their associated behavior and payloads, the API response time varied between 1 to 8 seconds. |