Re: [802.3_NGEPON] 802.3ca (100G-EPON) bi-weekly consensus - call for agenda items

["Salinger, Jorge" <Jorge_Salinger@xxxxxxxxxxxxxxxxx>]

Folks,

I mentioned the following during the meeting yesterday, but I thought I would send a note to make sure the point does not get lost in the conversation because I see comments from several people to the contrary . . .

NOTE: in all the discussion below I am ignoring PHY and MAC layer overheads to simplify the example calculations.

While it is absolutely correct that if you have a network with X capacity, you would not have service tiers that are rated at that X capacity. We covered this somewhat extensively during one of the Study Group meetings, which, to summarize, goes somewhat like this:

• Flagship peak tiers (these are the ones that the majority of the subscribers take) are generally scaled up to 1/3 of the X network capacity
• Bilboard peak tiers (these are the advertised highest tiers that very few subscribers take) are generally scaled up to 1/2 of the X network capacity

For example, 

• If the service tiers are 100 Mbps for most customers and 150 Mbps as the top speed that proportionally few customers subscribe, then we would engineer the network to support at least 300 Mbps of overall capacity
• Conversely, when a 10G-capable network is deployed (as we are doing now with 10G EPON), we would scale up to a 3G main speed tier for most subscribers and a 5 Gbps max speed tier for the lower penetration service.

If we need to have higher speed tiers, then we need to have higher network capacities. This is why we have DOCSIS 3.1 in the deployment phase and why we are convened at IEEE to develop NG-EPON.

As Glen described in the opening slides, all of which I agree with except for the bottom comment on slide #3 regarding the speed of development of the standard, we want to have a standard that is scalable over time, such that it can deliver more and more capacity for faster and faster speed tiers, without having to develop a new standard (until we get to the 100 Gbps network capacity range). 

As a specific example, referring to the above network design rule-of-thumb approach, with a 25 Gbps capable network we would have a flagship speed tier of about 8 Gbps and a billboard speed tier of about 12 Gbps.

So, what do we do when we need higher speed tiers than that?

The answer is not going to be to deploy more 25 Gbps wavelengths and equipment whereby any one ONU can only peak at 25 Gbps. This is what you get with tunable optics in NG-PON2, for example. The answer has to be that you deploy more wavelengths and equipment whereby a single (new) ONU that supports the additional wavelengths can send/receive data across the multiple wavelengths, having the ability to transport across the network more capacity than a single wavelength ONU would be able to.

For example,

• When we add a second wavelength and corresponding equipment we can have an aggregate network capacity of 50 Gbps, and thus be able to support a flagship speed tier of 15 Gbps (~1/3 of 50 Gbps) and a billboard speed tier of 25 Gbps (~1/2 of 50 Gbps)
• Similarly, when we add a 3rd and 4th wavelengths and corresponding equipment we can have an aggregate network capacity of 100 Gbps, and thus be able to support a flagship speed tier of 30 Gbps and a billboard speed tier of 50 Gbps

If we can’t do that, then adding wavelengths gives me the same outcome than splitting the service group into smaller groups of subscribers and using the same equipment. While I have to deploy more backhaul fiber to do that, or use a PON extender/mux, the equipment is the same, so I buy more of it, and it gets cheaper, as a result of which the cost of the additional backhaul fiber is at least in part overcome.

For example,

• The next gen of a 10G EPON is not a version of EPON that gives me two 10G EPON wavelengths, which would be the same as splitting the 10G EPON PON into 2 PONs with 1/2 of the subscribers
• By the same token, the next gen of a 25G EPON PON is not a PON that supports two 25G EPON wavelengths where ONUs can only support a maximum of 25 Gbps

Maybe this was all clear already. I know it is to us operators. We have been dealing with this same growth problem since the beginning, especially with DOCSIS. But, I get worried when I see statements like:

• “if we want to make a real low cost 100G EPON system,  from me, we should focus on how multiple ONUs with 25Gb/s  peak rate to achieve a 100G EPON system” from the Email from Dekun below; to me, this is not at all the goal of the scalable NG-EPON PAR and Objectives, or
• “I don’t mean I exclude the 100Gb/s peak rate ONUs, it can be applied for some very high-end users”, also from the Email below; it is not for high-end users that we eventually need 100 Gbps capable ONUs, but instead it is for the all of the customers as peak speed tiers continue to grow at the historical 50% year-over-year

I know that many people say things like “who could possibly need that kind of speed”. To that I always respond: “try to put your mind in the 2005 timeframe, when having a cable or DSL modem providing the capacity of 2 or 3 T1s to the home seemed incredible; would you have imagined then that you would have 2 or 3 DS3s dedicated to your home?” These are the flagship services that we offer today, and we already have 1 and 2 Gbps peak speed services!

Thanks!

Jorge

From: Liudekun <liudekun@xxxxxxxxxx>
Reply-To: Liudekun <liudekun@xxxxxxxxxx>
Date: Wednesday, February 17, 2016 at 4:20 AM
To: STDS-802-3-NGEPON <STDS-802-3-NGEPON@xxxxxxxxxxxxxxxxx>
Subject: Re: [802.3_NGEPON] 802.3ca (100G-EPON) bi-weekly consensus - call for agenda items

Hi Glen and All:

 I quite agree with Glen that “If 100G-EPON technology fails, It will be because of high cost”.

Let’s consider that why we choose the PON as the final access technology in the past.  It’s due to that it’s a point to multi-point system,   32/64/128 users can share one fiber , one OLT,  that’s the reason PON is a low cost system. 

 

When we consider 1G EPON, its’ 32/64/128 users sharing 1G bandwidth ,not a dedicating 1G bandwidth.  So it’s same for 100G EPON.  100G EPON should be 32/64/128 users sharing 100G bandwidth. But the absolute majority 100G ONUs don’t need to support 100G peak rate.

 

That’s the difference between a single wavelength TDM system and a multiple wavelength system.  For a single wavelength TDM system,  the peak rate of ONUs is always equal to the total system capacity .   But for a multiple wavelength system,  the peak rate of the ONUs is not necessary equal to the total system ! (The peak rate of a 100G EPON ONU can be only 25Gb/s)

 

The cost of ONUs are the majority cost of the total PON system,   if we want to make a real low cost 100G EPON system,  from me, we should focus on how multiple ONUs with 25Gb/s  peak rate to achieve a 100G EPON system.

 

By the way,  I don’t mean I exclude the 100Gb/s peak rate ONUs, it can be applied for some very high-end users,    but a 100G EPON ONU with only 25Gb/s peak rate should be able to satisfy the  major scenario in 100G EPON   , from the cost effective consideration .

 

For page 7,  I didn’t understand very well about what “100G ONU” really means,  if all the 100G ONUs means a 100Gb/s peak rate ONU in the slides(as they are shown in the figure),  I need to differ with this.

 

 

Best regards

Dekun Liu

____________________________________________________

Advanced Access Technologies Dept. 网络研究接入技术部

Huawei Technologies Co., Ltd. 华为技术有限公司
  Phone: +86 027-59267217  Email: liudekun@xxxxxxxxxx

 

From: Glen Kramer [mailto:glen.kramer@xxxxxxxxxxxx]
Sent: Wednesday, February 17, 2016 3:05 AM
To: STDS-802-3-NGEPON@xxxxxxxxxxxxxxxxx
Subject: Re: [802.3_NGEPON] 802.3ca (100G-EPON) bi-weekly consensus - call for agenda items

 

Curtis,

 

I'd like to discuss the general architecture approach. We had a number of presentations in Dallas leading towards this approach, but since the Dallas meeting was per-TF, we didn't make any formal decisions. In Atlanta, we started looking into various low-level details, but the big picture is left undefined. My slides are attached. All feedback is welcome.

 

Thanks,

Glen

 

On Mon, Feb 15, 2016 at 10:49 AM, Curtis Knittle <C.Knittle@xxxxxxxxxxxxx> wrote:

 

Dear Colleagues,

 

This coming Thursday, February 18, 11:30-1:00 Mountain Time, we will hold an IEEE 802.3ca 100G-EPON consensus building meeting. Please let me know by Wednesday noon (Mountain Time) whether you have requests for agenda time.

 

If I do not receive agenda requests by noon Wednesday, I will cancel the meeting.

 

Thank you!

 

Curtis

 

 

 

 

Curtis Knittle

VP Wired Technologies – R&D

CableLabs

desk: +1-303-661-3851

mobile: +1-303-589-6869

c.knittle@xxxxxxxxxxxxx

 

Stay up to date with CableLabs: Read the blog and follow us on Twitter

 

 

--

--------------------------------------

Glen Kramer

Broadband Technology Group

(707)529-0917