Why Content Providers don’t like Access Service Providers

Why Content Providers (Over the Top) don’t like Access Service Providers ?


Probably title of this post could be a ‘ Power of Access Providers ‘ or better ,  should be ‘ Why Some Content Providers don’t like Some Access Service Providers’. You will understand the reasons at the end of the post I promise.



I explained Content Providers , mostly known as OTT (Over the Top Providers) , Access Service Providers , Transit Service Providers and many other different provider business types in my Service Provider/Telco Training.


For those who haven’t taken the service provider/telecom training, let me briefly explain you what is Content Providers and what is Access Service Providers.


Content providers , such as Google , YouTube , Facebook , Netflix and so on provide a content (mostly they don’t generate the content) to the end users , which are called as ‘ eyeball ‘ in Interconnection business.


Access Service Providers provide a physical access , mostly known as last mile provider , through fixed or mobile connections.


For example , you might have xDSL , FTTx , Cable , BPL, Fixed Wireless , GSM, 3G, LTE connections from your service provider and that service provider is called as Access Service Provider.


Access Service Providers receive full Internet routing table from Transit Service Providers , but Transit Providers , such as Level 3 , TATA , and others (Tier 1 Operators for example) are out of scope of this post.


Traditionally , Content Providers , such as Google , YouTube (It is delivered as GCC – Google Global Cache, all my Service Provider Customer have GCC nodes in their network) , Facebook , install their servers directly inside the ISP networks. Content Providers install their servers in the IXP (Internet Exchange Point) and they do Open Peering through Peering Fabric with anyone else in the IXP.


But installing their servers in the Access ISP network , not in the IXP , bring Content Providers closer to the end users. Of course the total traffic of the ISP to that particular Content Provider should be sufficient, as those servers are costly , so server cost and the interconnection cost should be justified.


So far, nothing is complex I think.


But why Content Providers don’t or may not like Access Service Providers ?


This is related with data flow/type of data and direction of the data.  Let me explain.


End users mostly download the traffic from servers.


When you stream the video from Netflix , YouTube etc. there is a two way connection between end users and the servers. (Client – Server)


Amount of data being send from the end user to server to request the data (Video traffic for example) is nearly 20 – 30 times less than the data send from server to the end user.


So, traffic ratio from the end users (Access ISP) to the Content Providers , is not balanced  and 30 times more data  travels in the Access ISP Network. Which mean capacity need to be arranged to accommodate the required end user video traffic by the Access ISP.


So, if Access ISPs carry the 30 times more traffic than Content Providers, why Content Providers are not happy,  ?


Because, some ISPs are aware of this problem (Those who are from the Content Provider won’t see this as a problem and there are counter arguments , but I think this is a problem) and they would like to charge to the Content Providers. This is not related with Net Neutrality as well , in my opinion.



So, Access Providers want to charge Content Provider , because traffic is not balanced , and having Settlement Free Peering Interconnection just doesn’t make sense for some of the Access ISPs.


At the end , Access ISPs , have the pipe (physical or virtual) to the end customers. And for sustainability , for some Access ISPs, not providing a free connectivity to the Content Providers, is right model , in my opinion. Content Providers (OTT) is not regulated as ISPs and they don’t have to build massive infrastructure (Both CAPEX and OPEX) , but ISPs , they have to. This explains me why out of 10 largest business, 5 of them are Content Providers.






Comcast for example, (U.S Based Cable Operator – Access Service Provider) charges Netflix. Agreement between them is called ‘ Paid Peering ‘ .


So , Comcast is not providing full internet routing table to the Netflix but just the customer subnets, so it is like a partial transit service.


It is always business, and different business would defend the Interconnection model which make sense for their business.


If you talk to the  Access Provider, they will advocate Paid Peering , and strongly would defend traffic ratios. I have, end users , they mostly download , why I will carry 30 times more data then my peer, my peer should pay to me.


If you talk to the Content Providers, they will advocate Settlement Free Peering and strongly would be against to Paid Peering and they will even tell you that this about Net Neutrality. We shouldn’t pay , settlement free peering brings mutual benefit. But they just want more , in my opinion.






At the end, Access Service Providers have a power , because they have the direct connection to the end customers/users, and most of the time , in many countries , end users don’t have many Access Provider Options or they sign a long term agreement with one Access Provider , so they don’t have choice to change the Access Provider.


Last but not least, majority of the Access Service Providers, provide free interconnection (SFI – Settlement Free Interconnection) to the Content Providers (At least big content providers) , such as Google, YouTube , Facebook, Akamai and so on. But some Access Service Providers managed to charge for peering to the Content Providers , which I think , necessary and good for ISP sustainability.




5 Replies to “Why Content Providers don’t like Access Service Providers”

  1. I don’t understand your repeating comment regarding the access provider having 30x more traffic. That makes no sense to me. The content provider is sending the SAME amount of traffic to the access provider, just in OUTGOING traffic instead of INCOMING traffic the access provider sees the traffic as.

    Let’s say a Comcast user is pulling 50mb/s video stream from Youtube. Youtube sees 50mb/s OUTGOING on their transit connects and Comcast sees 50mb/s INCOMING on their transit sessions. It’s the exact same amount of traffic required on each end.

    1. It is not that simple. SFI works based on hot potato routing. So each peer sends the traffic to the closest exit. So receiving peer may need to carry the traffic longer in their network.In your example, Comcast may need to carry longer in their network. Thus, when content provider puts their nodes directly into ISP, you don’t have this problem.

      But not all content providers can place a node in each ISP, it is not economical if there is no enough traffic, that’s why they peer at IXP.

      Not all access providers ISPs allow every content operator to be in their POPs,DCs too , because charging is better than giving free interconnection.

      When two network peer , they want to peer in multiple location , because as I said in the beginning, peering works based on hot potato, every network does whatever is best for them, not for peer.

      1. Thanks for clarifying this. I work for an ISP with OTT peerings in its biggest POPs to offload transporting OTT traffic across the national backbone. This cost savings justifies the SFP arrangement, even when still accounting for additional space, power, cooling. I was wondering how your argument applied to the network I manage, but it is clear to me now the SFP does not make sense if you need to transport it across a backbone.

Leave a Reply

Your email address will not be published.

This site uses Akismet to reduce spam. Learn how your comment data is processed.