How to assess if a network provider can handle your traffic

Network capacity
5 December 2023

Understanding network capacity plays a pivotal role in ensuring seamless online experiences and making informed decisions about your online presence. In this blog, we’ll delve into whether a provider can handle your traffic and deliver the performance you need 

Table of Contents

Let’s dive into some basic concepts that are required to understand if a network provider can handle your traffic and deliver the performance you need.

Difference between real-time and cacheable network capacity

What is network capacity?

Network capacity refers to the maximum information transfer limit of a network at any given point. Network capacity can be further broken down into cacheable content capacity and real-time data capacity. 

Cacheable content capacity

Cacheable traffic consists of static resources that don’t change often, such as images, videos, CSS files, JavaScript and web pages. The capacity you use for cacheable content, which might be far away from the content’s region of origin is called cache capacity. 

For example, when you use a Content Delivery Network (CDN), you can put one copy of a piece of content on a server and then distribute the same copy 1000s of times throughout the entire network, throughout all caches of the CDN edge servers.

In some cases, you will encounter the term “deep cache” which is the same as regular cache, however, they are positioned deep in the targeted Internet Service Provider (ISPs). You usually only see big networks receiving CDN deep caches directly on  their network.

Real-time data capacity

Real-time data capacity refers to the ability of a network to handle and transmit data in real-time, without significant delays or latency. This is crucial for applications and services that require instant data transmission for multiple users, such as online multiplayer games or real-time communication platforms.

How to assess if a network provider can handle your traffic

Is the provider's network optimized for real-time data or cached data?

Whether a network provider can meet your connectivity needs depends on the specific use cases. For example, based on the type of customers the network has, you can determine the kind of traffic their network is optimized for. 

Look beyond the marketing numbers to understand the true value

One important detail to mention is that it is extremely difficult to tell whether the provider will be able to handle the traffic or not, and transparency plays a big role in this. When it comes to the major public cloud providers, many are not always keen on showing the real capacities they have; neither in networking nor compute, which creates operational issues for customers all over the world, as the information on the regional capacities is not disclosed, even if you are a long-standing customer with NDAs in place.

Martijn Schmidt

When showcasing network capacities, operators often combine deep-cache and real-time capacity, potentially confusing those unaware that real-time capacity holds greater importance for interactive multi-user projects like online multiplayer games and real-time communication.

Look into factors that affect your network performance and reliability

Understanding the components of network capacity is crucial because they impact the performance and reliability of your network. Network capacity isn’t a one-dimensional metric but a composite of various components and includes:

This reflects your network’s ability to exchange data with other networks at internet exchange points (IXPs). A higher IX capacity indicates a greater number of networks connected, ultimately leading to more efficient interconnectivity between networks. 

Private peering allows direct connections between networks, reducing reliance on transit providers and Internet Exchanges for further traffic decentralization, all while enhancing data transfer efficiency.

Transit purchased from third-party transit providers allows to carry traffic from one network to another, often across great distances or between different networks. 

Transport or transmission capacity is a means to an end since it allows networks to backhaul traffic to leverage their peering capacity in other metro areas for its delivery, making the network in question less reliant on third-party transit.

Qualitative differences

The mixture of external network capacity components can be different in every network. One network might have a lot of PNI capacity, a similar amount of IX capacity and some transit and no backbone capacity. Some networks are 100% reliant on third-party transit, making their networks fully dependent on third parties, meaning that they have no control over network issue resolution times, maintenance and/or outage reports

Additionally, being heavily dependent on transit providers could increase vulnerability to disruption as the network’s stability becomes contingent on the reliability of external transit providers. Meanwhile, the users would face higher latency due to the involvement of multiple intermediaries in data transmission.

Can you judge the reliability of the network based on network capacity?

In simple terms, yes, you can. If you can determine whether the network has a huge number of IXs connected, that usually means that the provider has diversified their connectivity and is “keeping its eggs in different baskets”, which reduces the so-called blast zone (size of negative service operational impact) making it less reliant on third parties.  

Conversely, if the network buys a lot of transit capacity instead of connecting to multiple IXs, it is very difficult to say whether that network is reliable or not, as the responsibility of providing reliable connectivity is pushed onto upstream network providers. This indicates that the network in question has very little control and visibility over any network faults. 

A network with more external capacity usually has more PNI capacity than capacity on the IXs. PNI traffic capacity is used to reduce the reliance on the IXs, just like IX capacity is used to lower the reliance on transit providers. That’s useful because it allows the network to maintain a higher level of independence from third parties, lowering the blast zone of any network faults.

Can you determine how big the network is based on the shown network capacity?

Network capacity doesn’t show how big the network is. Network providers can boast about a network that is many terabits, with multiple 100G ports ordered with upstream network providers, yet for a customer using a particular type of traffic, it’s nearly impossible to see whether the provider can handle the customer’s traffic or how efficient the network is in terms of routing. 

What you can do is get a limited idea on the size of the network from PeeringDB, where you can see content networks showing details on the amount of traffic being pushed through their network via IXs.

Note: Keep in mind that PeeringDB is a user-maintained database, therefore the information there might not be up to date. Yet it is commonly accepted by network experts to be by far the most convenient and publicly accessible data source. 

Another thing you could do is look at the IX capacities of the content network and get an idea about how big they are. If let’s say, a network is boasting about pushing many terabits of traffic via their network and they only have a few 10G ports with IXs, they are most likely a transit-heavy provider, making them heavily dependent on third parties. This signals limited control over their network, inconsistent connectivity performance and scalability limitations.

Many of the larger IXs around the world also publish the overview of IX members including the number and size of the ports, although members can opt out and not show any information about their presence in the IX.’s network capacity

When showcases its 23+ Tbps network capacity, it’s purely talking about the real-time data capacity, excluding any cacheable capacity. This shows the true value for interactive and multi-user applications like multiplayer games and real-time communication applications. 

With 9000+ direct peering connections we reach nearly all local ISPs’ “eyeballs” without using transit providers, ensuring minimal intermediaries and optimal performance for multi-user latency-sensitive applications. also provides full transparency on its network, by showing capacity split on its location pages, with the lion’s share of capacity split over IXs and PNIs, with relatively little reliance on third-party transit providers. This allows to be in full control over the network and for its users to understand how’s network works in detail. 

Additionally, owning and operating our own backbone enables us to identify and resolve network issues with greater efficiency and speed compared to other compute providers without a backbone. This backbone operation also facilitates more efficient data routing, resulting in lower latency and enhanced performance for our users.

In summary

Only looking at the network capacity number does not help you draw any concrete conclusions on the network’s size, reliability and the type of traffic it is able to handle. What is important is that you understand how network capacity is combined, what percentage of traffic is flowing through Internet Exchanges, private peering or third-party transit providers. With limited public information and in many cases unclear insight provided by the network provider itself, it is difficult to determine whether the network suits your use case without actually testing it. Pick your network provider carefully!

Main Take-Aways

The article helps demystify network capacity, distinguishing between internal and external components. It advocates assessing reliability through diverse connectivity and testing, cautioning against relying solely on capacity numbers.