Are IP/Optical Integration Initiatives Moving to Silos?

Will SDN finally enable the IP and optical convergence in the WAN that has eluded the industry for a long time?

After all, for how much longer can operators continue to suffer?

The two networks are built independently, are expanded independently and are maintained by two operations teams that work in complete silos.

And the result? Each layer is over-provisioned to cope with uncertain network demands and overprotected by using resources on each layer.

This is really painful.

We are talking about additional capex because of transponders and router ports -- the most expensive part of the core of the network.

And we're talking about huge opex, because of multiple teams handling the provisioning of a single service. Then there's the loss of time because multiple processes and teams are being deployed.

This raises a question: Can SDN come to our rescue?

An SDN controller that can decide on optimizing resources across layers, intelligently route traffic on the low-cost layer and enable an operator to provision services, whether IP or optical, from a single management system would indeed be a welcome addition to the network of an operator.

That's why the SDN use-case for multilayer IP/optical optimization in the WAN has received a lot of industry attention.

But for such IP/optical integration to happen, the industry must collaborate. That means the broader participation of all stakeholders, including the IP/MPLS vendors, the optical vendors and the operators.

Anything less would be a silo or less effective initiative and would result in a strategy that represents the voice of one segment of the industry.

Like the one from the Transport SDN group of Open Networking Foundation (ONF): That represents the voice of the optical transport segment of the industry, minus IP. And understandably so, because this group is chartered to bring SDN to Layers 0, 1 and 2 (transport), and not Layer 3 (IP).

So even though there is a clear use case of IP/optical integration in the ONF white paper on Optical Transport Use Cases, the efforts to bring such optimization would only be effective if either mainstream IP/MPLS vendors join the group or close collaboration happens with them.

But ONF is not alone. A close look at the initiatives in the industry reveal the dispersed and disintegrated status of the IP/optical convergence initiatives (see the table below).

Table 1: IP/Optical Convergence Initiatives

Initiative  Lead members Goal Why this is a silo
IP/optical initiative of ONOS AT&T, NTT, Ciena IP/optical Integration Seems a very good initiative with a clear strategy. However, it is missing broader industry participation, especially IP/MPLS vendors. AT&T is the most active member.
IP/optical initiative of the Telecom Infra Project (TIP) Facebook, DT, Juniper, ADVA, Coriant IP/optical integration Unclear strategy as still new. Needs broader industry participation.
Optical Transport Group of ONF All traditional optical (Huawei, Ciena, Nokia etc.) Transport SDN. It has a use case for IP/optical integration. Optical vendors' silo.
NSP from Nokia* Nokia proprietary IP/optical integration Single vendor silo.
nlight System from Cisco* Cisco proprietary IP/optical integration Single vendor silo.
* Nokia and Cisco included owing to a clear IP/optical strategy and available commercial products.
'Broader industry participation' in this instance means significant participation of all three main groups: traditionally optical (for example Huawei, Ciena, ZTE, Nokia, etc), traditionally IP/MPLS (such as Cisco, Juniper) and operators.

The same goal but six independent initiatives!

Wouldn't it make sense to have more focused efforts on IP/optical integration? Wouldn't it auger well if there are just a few open interfaces for SDN rather than many?

Even though some vendors would boast of open interfaces on their SDN platforms, this is not enough. Firstly, not everyone supports multiple open interfaces. Secondly, open interfaces do not always mean plug-and-play systems: There will always be a need for some scale of interoperability tests between the vendors.

Therefore, buying a solution of one silo would mean the risk of interoperability issues with a solution from another silo.

If that happens, the operator would again be stuck in a silo. This time, it would not be a silo between an IP and optical network, but a silo of a certain group or a vendor, and that even be worse!

The industry needs collaboration and integration. Too many initiatives may de-focus the industry from achieving the ultimate goal of IP/optical integration. Therefore I ask for you, the industry, to share your opinions here (using the message board below) on how this collaboration can be enabled for such an important issue.

— Faisal Khan has 15 years' experience working with a range of network operators across the Middle East and Africa region, including one where he is working now: The views expressed here are his own and do not reflect those of his employer.

Page 1 / 2   >   >>
aazizk 8/22/2016 | 1:10:22 AM
I think it is more of a market share game Dear Faisal,


It is a nice topic that you have started and I will try to contribute as per my understanding.


What I see, while sitting in different places in vendor and operator sides; is that this is all a game of market share, each vendor sees this integration of IP and Optical as a loss of his market share in his domain because the small operator who do not have skilled resources at their hands play in the hands of vendors.


The game changers can be either big vendors like Huawei who is present in both IP and Optical domain or the big operators who can force vendors to their own terms.


Untill and unless some big players really focus to bring this change it will take a lot of time to come.
OriG464 8/4/2016 | 4:53:10 AM
Re: IP Optical integration is ready, here and now While standards are improtant, the industry is realizing that waiting for them means missing out of the huge value that IP/optical SDN integration can bring. Look at the Google's of the world: they just make stuff happen, learn on the way, and by the time standards are there - they are well ahead of the traditional network operators.

Check out my blog entry on the topic http://sedonasys.com/index.php/resources/blog/210-the-road-to-sdn-one-step-at-a-time for a possible smooth evolution path that Sedona's technology enables.
batye 8/2/2016 | 7:48:59 PM
Re: Double-vendor lock-in / standards @Faisal Khan, yes same is on my end I could not agree more as model and approach is changing how I see it ...
Faisal Khan 8/2/2016 | 12:45:44 AM
Re: IP Optical integration is ready, here and now @CEO73432

I agree with you. Although standardization would be necessary to streamline the efforts the industry is taking at the moment.

Faisal Khan 8/2/2016 | 12:42:43 AM
Re: Double-vendor lock-in / standards @Director82455

1. I agree with you; Also,there is an ongoing work in IETF to have two two controllers exchange topo information through YANG models so even though the optical and IP network elements are not integrated at NE level, there is still a way to share TE information and have an SDN contorller act accordingly.


2. Yes,it is clear that owing to too many standard bodies, it is difficult especially for operators to join all of them, thats why the industry needs more concentrated and integrated efforts instead of silos.
CEO73432 8/1/2016 | 3:49:05 PM
IP Optical integration is ready, here and now Great analysis Faisal !

Network operators who want to enjoy the benefits of IP-Optical integration don't have to wait for standards to cristalize, especially as there quite a few of those initiatives.

The road to SDN one at a time explains how this can be done.

Sedona Systems is already doing iP and optical L0-L3 integration on carrier's networks in a multi vendor and multi domain environment.
Director82455 8/1/2016 | 10:44:17 AM
Double-vendor lock-in / standards Hi,

There are 2 comments from my side:

  1. The single vendor solution is the most difficult direction at the market at the moment for operators as especially small to medium operators fear a lock-in situation as they may have only single vendors in the network. A vendor specific SDN solution may look interesting at the beginning but it forces the vendor lock-in on the HW as well as on the SW orchestration layer. Bringing in a second vendor in the network on the HW layer may be impossible as it requires the integration into orchestration and application layer.

  2. Driving standardization through the standardization organization faster – is this really realistic? Most of the guys in this groups are from the vendors as a lot of the big operators reduced the number of people in the group or even left the organizations, but as of the political issues of standardization, the operators are the most important guys to drive standardization forward ...
marjorysy 8/1/2016 | 4:43:36 AM
Thanks thanks for the share. mind-boggling :)
dwx 7/29/2016 | 11:15:47 AM
Re: well defined Faysal GMPLS never amounted to what it was supposed to with regards to interoperability between layers or platforms.  GMPLS UNI is a good idea, but lack of provider interest and vendor adoption led to its demise.  There are closed vendors like Nokia (ALU) and Cisco (nLight) using GMPLS UNI with some other augments, but it's a closed system.  Thus, using SDN and having controllers speaking to eachother via open APIs using standard data models is more feasible today.  No one uses OpenFlow, not sure where that became a "standard."  

Vendors are being forced to IP/Optical integration for datacenter interconnect by the large content players.  MS wanted to simplify their network by looking to move to 100G DWDM QSFP28 optics via InPhi, so they can use packet/optical integration in any device.  Cisco, Juniper, Arista, etc. are all building DWDM blades for their IP gear based off ACO to support these needs and simplify the network by eliminating another active component layer.  The future though IMHO is going back to DCO like the InPhi, so the packet HW only needs to have software support for the DWDM component.  
Faisal Khan 7/29/2016 | 3:03:49 AM
Re: Comment on ONF transport @dudshir, would you really need such a big body to develop just BoD through Transport SDN ?

I think the community should solve the problems  first ( that costs money to SPs) , before thinking about monetizing through BoD.

This is high time the Transport SDN of the ONF should broaden the scope and bring the larger IP community to its fold instead of the silo work
Page 1 / 2   >   >>
Sign In