Comcast: First let's talk about Comcast. Articles across the blogosphere keep referring to the time when Comcast was "blocking" peer to peer traffic. While this is technically correct, it does not really portray what was going on.
Comcast found that a small percentage of their users were consuming a very high percentage of the network bandwidth, and they were predominantly using peer-to-peer transfer protocols like BitTorrent, swapping music or video files. They took the view that users consuming high volumes of traffic was fine unless it impacted their neighbors. The cable plant shares bandwidth between a group of neighbors. In a typical scenario there might be 6 Mbps of uplink bandwidth that is shared by 100 or 150 users, each of which is allowed to use as much as 1 Mbps of upload bandwidth on a momentary basis. Obviously not all the users can use 1 Mbps concurrently, but in most cases this is not a problem. Uplink bandwidth is typically small unless you are uploading pictures, updating websites, or transferring music using BitTorrent.
If six people, in the scenario above, are all using BitTorrent, it is possible for them to fill the uplink pipe, and then cause performance problems for the other 144 neighbors using the same bandwidth.
Comcast chose to alleviate this problem by slowing down BitTorrent for heavy users. BitTorrent does not consume all its bandwidth with one TCP connection, but instead opens many connections to many different PCs around the world, and sends off pieces of a file to each one. Comcast monitored those connections, and limited the number of links a PC could open by killing TCP connections beyond that limit. So BitTorrent was being slowed down by blocking some of the many threads it opened. But BitTorrent was still operating. Most of the music transfers got through, they just took a little longer.
After receiving criticism for this approach, Comcast changed their methodology. The new approach monitors the utilization of each user (total bandwidth consumed over a measurement period) and then lowers the priority of all traffic from that user if the uplink is congested and if the user has been a heavy user. Priority is reset to normal after the user demand or the total uplink demand drops to a less congested state. The effect of this approach is that heavy users can use the network when they are not in contention with their neighbors. When they are in contention with neighbors, the lower volume neighbors are given priority, but the heavy user is still given every possible byte of available bandwidth to continue to upload their files.
The same algorithm applies to downlink bandwidth, except the bandwidth numbers are much higher. Management of uplink and downlink bandwidth is independent.
|
logon-to-comment
|
Quick View | Full View | 0 Comments |
This is a public forum. UBM TechWeb and its affiliates are not responsible for and do not control what is posted herein. UBM TechWeb makes no warranties or guarantees concerning any advice dispensed by its staff members or readers.
Community standards in this comment area do not permit hate language, excessive profanity, or other patently offensive language. Please be aware that all information posted to this comment area becomes the property of UBM TechWeb and may be edited and republished in print or electronic format as outlined in UBM TechWeb's Terms of Service.
Important Note: This comment area is NOT intended for commercial messages or solicitations of business.
Focus: SIP Trunking
Featured Article: Under the Hood of Microsoft Communications Server 14
· CONTACT CENTERS Sponsored by
Webinar Replay: Contact Centers & Migration to IP
Recent blog: When Preparation Meets Opportunity in the World of Customer Care

No Jitter for quick reference. It's easy and free! CLICK HERE to register and get started!

