How does PRTG scale? (EN)
Hello everyone, I’m Nina and I work for Paessler’s Presales Team.
In this short video, I would like to talk about how PRTG scales. If you are planning a large installation, or want to scale and speed up your existing installation this is just the right video for you.
We want to make you aware of the various factors that contribute towards the performance of a PRTG server and how to get the most of it.
Scaling PRTG means to implement a large PRTG installation in the best way possible. Scaling becomes more important, as networks become bigger and bigger. Existing and potential customers with large networks, who want to upgrade, often have performance issues.
When you have a large installation, we do not restrict you from using a large number of sensors. Depending on your environment, performance can be acceptable with a higher sensor count. Since each environment is unique, we cannot predict the behavior of your PRTG installation. We want to show you several ways to speed up PRTG.
It is important to explain how PRTG scales: whether you run PRTG on a virtual environment, or on dedicated hardware, you always start by scaling vertically. This means you add sensor after sensor to your installation. Once you start noticing that PRTG slows down you have different options:
One option is to review the state of your current virtual server. You should either allocate additional virtual cores and memory, or migrate to a server with less contention and newer hardware.
Check if the following factors are balanced: Short scanning intervals lower the performance, so try using longer scanning intervals. It is also recommended to use lightweight sensors, whenever possible. Heavyweight sensors like flows, WMI and packet sniffer should be used with care. They need more resources than a ping or SNMP based sensor.
Furthermore, use remote probes to distribute the pulling workload. To decrease the CPU overhead movie sensors to remote probes rather than running them on the local probe. Additionally, reduce the amount of auto-discovery you use, as it generates a significant load on your core server. You can also deploy an additional core server, which is called horizontal scaling. This again works for both installation types, VMs and physical servers. If you are running PRTG in a virtual environment and do not wish to deploy a second core server, you also have the option to improve the speed of your installation by moving to dedicated hardware. Check out our Knowledge Base guide on how to migrate your PRTG configuration. If you decide to go ahead with a physical server, we recommend using Multi-Core, Multi-Threaded CPUs, or Multi-Processor Systems. You should also use a Fast Disk Sub System and install additional RAM Memory on the PRTG core server. It is best to use a 64-Bit operating system. There are other PRTG features that you should not use excessively, due to their impact. These features are clustering, extensive reporting, creating thousands of maps and the overuse of the API. Furthermore, you should always keep your PRTG installation up to date. This does not only fix bugs, but also helps your installation to run smoother. Last but not least, a stable network connection is needed. Many of our customers run huge installations on a single core server, because they managed to find the right balance of all the previously mentioned factors.
So, if you still have questions on scaling or planning your PRTG installation, just contact us at [email protected] We are always happy to help. Thank you for watching, goodbye.