NFV (Network functions virtualization)

NFV to Trigger Cisco Buying Binge?

This week's report from MKM Partners on AT&T's future buying habits appears to paint a bleak picture for Cisco in the future realm of virtualization, but its overall view of that giant company's future is actually much more measured -- and includes the very likely possibility that Cisco, and possibly Juniper Networks, will be making major acquisitions in the near future to buttress their positions. (See: MKM: Cisco Biggest Loser in AT&T SDN Plans.)

As a result, the industry could see major consolidation affecting optical transport vendors such as Infinera Corp. and Ciena Corp.

For the near-term, the report notes that AT&T Inc. (and by extrapolation other major carriers) "will continue to buy routers from Cisco, Juniper Networks Inc., and Alcatel-Lucent and Optical transport equipment from Ciena in 2014-2016, and likely well beyond."

After that, MKM Partners expects the full convergence of network layers 0 through 3, as network gear compliant with network functions virtualization (NFV) is deployed, enabling data and optical equipment to merge into virtualized wide-area network routing.

"We expect that Juniper and Cisco will continue to sell WAN packet networking equipment to AT&T well into the future, but they may need to add ultra-high capacity coherent Optical transport to their product capabilities," MKM notes. "For this reason, we view best-of-breed Optical transport vendors like Ciena and Infinera as potential acquisition targets for the larger Router vendors such as Cisco and Juniper over the next several years."

We know that NFV and its partner in virtualization, software-defined networking (SDN), will be major disruptors in the public network, and this report is one interesting glimpse into how that disruption is beginning to play out.

— Carol Wilson, Editor-at-Large, Light Reading

Page 1 / 3   >   >>
Dredgie 1/7/2014 | 12:05:05 PM
Re: We may see an industry reshaped >> Heya, Gabe - 14 drafts in progress (covering a large swath of issues) prior to achieving WG status on Dec 20th, following a very active charter discussion on the boards. They nearly made WG status in Vancouver (the last IETF meeting) based on the progress, but there was one last BOF instead, as the charter text was cleaned up.

I'll bow to Tom regarding the relative complexities. While the group calls-out the need to support non-virtualized functions along with virtualized, there are defo many synergies. This WG was born from NFV, after all. In the hierarchy, it's the next layer up, I think - where forwarding graphs group VNFCs (i.e. a SIP Router, Charging Trigger and an HSS Mirror) into VNFs (an IMS Core) and the service function chain then includes this VNF with an SBC, VoLTE TAS, and the like.
TomNolle 1/7/2014 | 8:30:56 AM
Re: We may see an industry reshaped Not necessarily.  The "service chaining" use cases talked about at the ETSI ISG level are linear data-plane-coupled elements because they are hosted equivalents of the connected devices from which the VNFs derived.  There's nothing complicated about saying that the goesouta in this box connects to the gosinta in that one, and if you simply translate functionality to hosted form then the connectivity model is the same.
Gabriel Brown 1/7/2014 | 8:21:51 AM
Re: We may see an industry reshaped Dredgie -- what's the status, roughly, of the Network Service Chaining work in IETF?
Gabriel Brown 1/7/2014 | 8:19:43 AM
Re: We may see an industry reshaped I'm not sure I agree that it is a trivial sub-problem. A service chain (or forwarding graph in NFV-speak) requires an information model and a way to communicate processing requirements to the software and hardware modules that make up the infrastructure. This is not a million miles from what NFV orchestration is about.

How much the two actually overlap, I dont know.
^ip4g^ 1/7/2014 | 1:50:28 AM
Re: We may see an industry reshaped tyre can be tire..  but all of tire cannot be tyre
TomNolle 1/6/2014 | 5:26:18 PM
Re: We may see an industry reshaped Maybe I'm wrong, but SFC seems to me to be nothing but another way of talking about service chaining, which is a relatively trivial sub-problem within the scope of NFV.  I don't see any progress on broader issues of modeling and management, and a forwarding graph is useless except to drive tunneling connections, which most current logic doesn't expect, so it wouldn't work unless you wrote VNFs from scratch.
Dredgie 1/6/2014 | 4:21:39 PM
Re: We may see an industry reshaped >> I dunno, guys – I think that was decided way back in 1783... and I'm not talking about the RFC. I can't believe I'm having to defend my adopted home country, here, Tom :-)

Anyway - back the topic: With the intensified focus on simplified operations, I'm increasingly looking at the work being undertaken in the newly formed IETF SFC (NSC) working group. While it was always interesting in terms of augmenting / being a natural extension, I'm seeing the real potential for it to trump NFV in terms of importance. Although the charter and goals are demanding (new data plane encapsulation formats, contextual continuity in the form of metadata, and the like) that seems to be the MO in these days of massive architectural overhauls (ref. the subject of this post thread), so why not!
TomNolle 1/6/2014 | 9:14:13 AM
Re: We may see an industry reshaped Low-class will always overrun high-class; majority rules!
Gabriel Brown 1/6/2014 | 8:41:48 AM
Re: We may see an industry reshaped Virtualised or virtualized?

Basically, British English has been overrun by American English. Oh well.

Fair point on NFV-washing.
TomNolle 1/6/2014 | 8:36:25 AM
Re: We may see an industry reshaped I think that the tyre/tire thing is a question of where you are with respect to centre/center!

I'm all for useful experiments, and I'll concede that there may be some experiments that could be conducted with respect to NFV right now, even lacking a unified operations model in which to contain it.  However...I think that the great majority of NFV announcements are pure fluff in that they don't actually prove anything useful.  For example, we can run virtual routers/switches on servers or in VMs now.  Do we say that's NFV simply because such a step is a logical part of NFV deployment?  In my view, NFV is "advanced" by advancing something that contributes to its long term value proposition or answers a question related to that proposition.  I'd like to see more discrimination applied to NFV claims so we weed out the things that are just NFV-washing.

Example:  We need to understand what an NFV plantform has to be able to do in order to host virtual functions.  If we run a series of tests designed to determine how various hardware components and software tools can accelerate the data plane or can improve availability, then we're answering an NFV question.  If we add to that the exploration of how such a hosted function might be managed at a lower cost than the current physical devices are managed, then we're directly addressing the value proposition for NFV.  If we run a firewall in a VM and say it's NFV, we're blowing smoke.
Page 1 / 3   >   >>
Sign In