How does PTA Agent deployment on the domain controller affects its performance?

cancel
Showing results for 
Show  only  | Search instead for 
Did you mean: 
PAM Self-Hosted Articles
3 min read
L0 Member
No ratings

Question

How does PTA Agent deployment on the domain controller affects its performance? Is there a limit (or guideline for a maximum) of events per time unit that the PTA Agent is able to process? And how much resources would it be consuming at that limit?

Answer

The DC Agent is an ultra-light agent on the Domain Controller that taps its traffic and detects Kerberos attacks.
It’s impact on the Domain Controller server is marginal, it consumes only 0.4% CPU on 4 cores system when processing 10,000 packets per second!

When we developed the agent, our goal was to have minimal performance impact, and CPU footprint over the DC. We aimed to avoid high resource consumption of the service both in average and also in terms of picks in the services resource consumption.
As another protection layer, we developed the option to enforce the service resources, meaning he agent have the ability to monitor the machine's and application's different resources (CPU /Memory etc.. ) and in case we recognize that the service resources exceeded a certain threshold, we stop it.

Please find the benchmark results for the PTA Agent in our online documentation through the link below:
https://docs.cyberark.com/Product-Doc/OnlineHelp/PAS/Latest/en/Content/PTA/Appendix-D-Benchmark-PTA-Agents.htm

Issue

Dear Support, Our customer, the National Police, have purchased domain controller protection licenses, however, they have concerns regarding installing the PTA Agents on the domain controller. Therefore, I have three questions: Background information: When running a PoC with a competitive tool, Microsoft ATA, it was discovered that the setup with deploying agents on domain controllers was not sustainable, because of the enormous amount of events and the resources that it took to process these, which it took away from the domain controllers. They needed to deploy the network sensor setup instead, to ensure that their domain controller services would remain available. Questions: 1. Is it possible that the PTA Agent deployment on the domain controller could cause similar performance issues on their domain controllers, because it would take a lot of resources to process a significant amount of events? Is there a limit (or guideline for a maximum) of events per time unit that the PTA Agent is able to process? And how much resources would it be consuming at that limit? 2. Is there a limit (or guideline for a maximum) of events per time unit that the PTA Network sensor is able to process? Would this setup be advised should they expect performance issues on the domain controller? 3. Do network sensors in any way provide a limited or less accurate (or perhaps an even better) capability to detect any of the kerberos attacks? As far as I can tell based on the documentation, the network sensors and PTA agents should have the exact same capabilities, but we only have experience with implementing and managing PTA Agents, so we would like to have that confirmed. Many thanks in advance. Kind regards, Paulina

Environment

PTA Network Sensor
Rate this article:
  • 1253 Views
  • 0 comments
  • 0 Likes
Article Dashboard
Version history
Last Updated:
‎12-17-2020 04:39 AM
Updated by: