AppNeta Capacity Measurements

Products

• AppNeta

Issue/Introduction

AppNeta Network Paths are able to get capacity measurements of the link between itself and the endpoint, even for single-ended paths (ICMP). 

How does it work when compared to traditional speed/capacity measurement tools?

Resolution

Traditional methods of measuring Capacity/Speed would require a client and server at each end and flood the network with traffic. PathTest is our traditional capacity measurement validation tool within the AppNeta Product, using UDP you would send a specified amount of traffic or maximum possible traffic to the target monitoring point.  This method works by flooding the network with the desired amount of traffic and gathering the resulting numbers of how much traffic was able to be sent.

 

AppNeta's Delivery Network Paths differ in the above approach in order to conserve network data consumption while still providing accurate numbers for most scenarios.  Both Single and Dual Ended Network Paths use the same method for measuring capacity metrics with packet dispersion.  As an example if two packets are sent on the network to a target, we are looking for the packet dispersion time between the last byte of the first packet and the last byte of the second packet.  Noting that the above scenario uses 2 packets, actual capacity measurement tests consist of roughly 15-50 packets depending on the conditions detected.

• Total Capacity is measure by dividing the Packet Size (in bits) by the Dispersion described above in seconds.
• Available Capacity is measured using the average dispersion of a series of packet trains taking any lost packets into consideration when it's calculated.
• Utilized Capacity is the Total Capacity minus the Available Capacity.

 

Generally speaking this measurement method only becomes problematic in scenarios where the ISP is using Rate Limiting methods which only trigger above a certain utilization threshold.  Meaning it would be possible for us to see the underlying connection, for example the underlying connection is 1Gbps but the customer is only provisioned 100Mbps, as our Capacity Monitoring is typically less than 100 Kbps we may not trigger the ISP Rate Limiter and as a result our capacity measurements reflect the full underlying 1Gbps connection.  
Support does have options available to assist with tackling this limitation by increasing the number of packets sent however to do so requires the customer to open a support case.

 

Measuring Capacity with the above methods can be done with ICMP or UDP, though UDP will provide overall the most detail.  As ICMP is designed to reach it's target and return to the source it works well with our methodology of measuring the dispersion between the first ICMP packet and the last because the packets still need to traverse the network to the target and back to the MP.

 

It's also important to understand that in our Monitoring, Capacity is not the same as Bandwidth.  Our measurements are end-to-end network layer measurements whereas Bandwidth typically refers to the physical interface capability.  As a result our measurements automatically take into account network level overhead from networking protocols such as L1 + L2, TCP overhead and so on.

Additional Information

The most detailed answer can be found in our TechDocs Page regarding Capacity Measurements as it breaks down the math further as well as other metrics: Capacity Measurements