|
Posted on 07.10.06
This article by Dan Parsons, who is director of marketing at BroadLight, details the advantages of GPON over EPON. One advantage that is emphasized is the bandwidth capacity to handle multiple HDTV streams while still having sufficient spare capacity for Internet service for subscribers: For example, with 45 SDTV and 5 HDTV channels on the PON, EPON can service up to 135 subscribers with high-speed Internet while GPON can support up to 413. However, as the video content goes to 100% HDTV with 50 video channels on the PON, EPON’s downstream capacity is exhausted while GPON can still guarantee high-speed Internet service to 278 subscribers. I would think this would be more of an advantage in the US with multiple HDTV sets per residence (in the near future at least) compared to Asia, and China in particular, where HDTV penetration would be lower. The second advantage that Dan Parsons talks about is GPON’s more efficient quality of service QoS mechanisms. Filed under: Technology Comments: 5 Comments »RSS feed for comments on this post. TrackBack URI Leave a commentLine and paragraph breaks automatic, e-mail address never displayed, HTML allowed: |
Clearly the writter has never actually experienced IPTV in a large scale commercial deployment.
Firstly, HDTV channels in the US (or anywhere for that matter) do not stream @ 18Mbps. Typical HDTV in the US streams at about 10Mbps. The Superbowl broadcast was streaming at only 15Mbps. In AsiaPac, streams are typically about 12-14Mbps.
Secondly, I have worked with delivering IPTV solutions for over 5 years, and even with a full channel lineup, I have never seen all channels being used at the same time, across the entire network, let alone a 32 way PON.
Even working with MDU installations (such as hotels with 200+ rooms) the typical number of channels used on a single PON is about 35% of active channels. This actually decreases in percentage as the number of available channels increases.
Even with the addition of PiP, PVRs, and multiple streams within the home @ the same time, this MAY push the number of different channels up to 50%.
So with a 50 channel line up, typically only 15 channels would be in use at any one time. Even if these where all HDTV, this would be only 180Mbps. If we increase this to 50% channel useage, this is still only going to be 300Mbps.
This still leaves 600Mbps for an EPON network, and using the Author’s 20:1 contention ratio (which is about 1/5 of the typical contention ratio used in carrier/isp networks today - its normally 100:1), this equates to 120 active users.
Hang on, there is only 32 splits on the PON, so claiming support for more than 32 active users @ 100Mbps each is irrelevant anyway. And we also know that not every user is active at the same time. Typically only 1 in 4 users are actually logged on to the internet at any one time.
Comment by Stephen Davies — July 12, 2006 @ 7:32 am
Can we please stop arguing amongst the industry as to which technology is better (GPON, EPON or Active Ethernet) and focus on that fact the FTTH (what ever the technology) is a neccessary infrastructure for the 21st century.
They each have their pros and cons, but the last mile access speed is not the limiting factor in bandwidth, its the backhaul!
Comment by Stephen Davies — July 12, 2006 @ 7:36 am
Good article. Bandwidth requirements for HDTV etc. may be too pessimistic in terms of what is available with MPEG4. But the key point about QoS is right on. The need for QoS in 3-play is one of the main reasons why GPON is selected by US carriers. GPON’s ability to have granular bandwidth allocation for a given ONT/customer with multiple services is key in a shared medium like PON. Unlike switched Ethernet, in PON (any PON) 16/32/64 customers share the same fiber and each customer needs to have QoS guarantee and a fair allocation of bandwidth for multiple services. This requires mapping the Ethernet layer CoS (802.1p) to PON QoS and bandwidth allocation scheme which is what GPON does. It does it very effectively because it uses embeded bandwidth allocation maps for up to I think 4K allocation IDs per PON which gives you plenyt of granularity without any additional overhead. With EPON, the number of Logical Link IDs (LLIDs) supportable in a PON without too much overhead becomes a problem.
Comment by SR — July 19, 2006 @ 10:33 am
QoS at the edge is overrated. The complexity that results negates the benefits it brings. Much easier to fatten the pipe and forget about the problem.
Even if you wanted QoS, nothing prevents you from using the 802.1p bits to drive variable bandwidth allocation over E-PON.
All of this nonsense just adds latency unless you make it even more complicated.
Just make it a fat pipe. Latency and QoS problems solved. If a user abuses this, cap his ass.
Comment by Andrew Schmitt — July 20, 2006 @ 8:42 am
QoS at the edge adds cost and latency to your network.
The FTTH market is about driving the cost down.
Comment by Stephen Davies — July 23, 2006 @ 5:46 am