This should be tested on the Dev Preview release.

Goals

• Support at least 20K NetFlows per second
• Ensure all UI response times are reasonable
• Document behavior when system is overwhelmed
• Take measurements (see below)
• Come up with general formula and rough calculation on resources needed for various scenarios

Hardware and Software Consideration

These are the different attributes to consider.

• CPU: 1 core, 2 cores, 4 cores, 8 cores
• Memory: 0.5 GB, 1 GB, 2 GB, 4 GB, 8 GB, 16 GB
• Storage: 20 GB, 200 GB, 500 GB, 1 TB, 10 TB
• Bandwidth: 10 GBps, 100 GBps
• Sampling rate: 400:1, 100:1, 1:1 (no sampling)
• Nodes: 7 (25th percentile), 9 (median), 126 (75th percentile)
• Type of traffic: breakdown of web (80/443), database, video, mail

Testbed and Measurements

1. (Typical) 2 cores, 1 GB RAM, 200 GB storage, 10 GBps bandwidth, 9 nodes
   1. 100:1 sampling, medium traffic
   2. 400:1 sampling, medium traffic
   3. 1:1 sampling, low traffic

1. (Min) 1 core, 0.5 GB RAM, 20 GB storage, 10 GBps bandwidth, 7 nodes
   1. 100:1 sampling, low traffic
   2. 400:1 sampling, medium traffic

1. (Max) 8 cores, 16 GB RAM, 1 TB storage, 100 GBps bandwidth, 50 nodes
   1. 100:1 sampling, medium traffic
   2. 400:1 sampling, high traffic

For each scenario, measure the following:

• Maximum NetFlows per second
• How many days/hours of storage before it reaches capacity?

Other questions:

• What resources are needed to reach 20K NetFlows per second?  50K NetFlows per second?
• How much storage is used for 1K NetFlows per second?  Is storage usage linear?