跨不同安全域的LoadRunner负载测试性能差距测量问询

Hey there, let's walk through practical ways to quantify the performance impact of VDI access and frequent token/authentication operations in your setup. Here's a step-by-step approach tailored to your scenario:

1. Establish a Baseline (No VDI/Auth Overhead) #

First, you need a benchmark of the system's raw performance without the extra layers:

• If security policies allow, temporarily set up a direct access path to the target VM (e.g., move a test instance to the same OU as your LoadRunner servers, or open limited test-only connectivity).
• Record a LoadRunner script that focuses only on core business operations—skip VDI login, token fetching, or refresh logic entirely. Keep it tight to the actual workflows you're testing.
• Run this script across multiple load levels (10, 50, 100 concurrent users, for example) and log key metrics:
  • Average/95th/99th percentile response times
  • Throughput (requests per second)
  • Resource usage on both LoadRunner physical servers and the target VM (CPU, memory, network)

2. Run Full End-to-End Tests (With VDI & Auth) #

Next, test the real-world scenario that includes all the extra steps:

• Record a complete script that mimics exactly how users interact: VDI session initiation, token acquisition, core business actions, and token refresh/renewal (follow the exact token expiration logic used in production).
• Use the same load levels as your baseline test to keep comparisons apples-to-apples.
• Log the same metrics as the baseline, plus additional data from the VDI server (session count, resource usage) and any auth service you're using.

3. Isolate Individual Overheads #

To pinpoint which component is causing the most drag, test VDI and auth separately:

• VDI Session Overhead Test: Record a script that only handles VDI login, session setup, and logout. Run it at varying concurrencies to measure how long session establishment takes, and check for bottlenecks on the VDI server (e.g., max concurrent sessions, CPU/memory spikes).
• Token/Auth Overhead Test: Record a script focused solely on token fetching, validation, and refresh. Simulate the exact frequency of auth operations your users face, and measure the latency of each auth step along with resource usage on the auth service.

4. Compare & Analyze the Data #

Now it's time to crunch the numbers:

• Calculate the delta between your baseline and full end-to-end tests:
  • Subtract baseline response times from end-to-end times to get the total extra overhead from VDI + auth.
  • Compare throughput drops—if throughput is significantly lower in the full test, that means the extra layers are limiting how many requests the system can handle.
  • Check resource usage differences: If the target VM's resource usage matches the baseline, the bottleneck is definitely in the VDI or auth layers. If LoadRunner servers are spiking, the extra network traffic from VDI might be taxing them.
• Use LoadRunner's Transaction Breakdown feature to split out the time spent on each step in your end-to-end script (VDI login, token fetch, business operation). This will show you exactly what percentage of total latency comes from each component.

5. Validate with Controlled Variables #

To confirm your findings, tweak one variable at a time and re-test:

• Try extending token expiration times (if possible) and re-run the full test. If performance improves noticeably, you'll know frequent token refreshes are a major contributor.
• Enable VDI session reuse (if your VDI platform supports it) and measure if that cuts down on session setup overhead.

内容的提问来源于stack exchange，提问作者thepen