SDN architectures

One Problem With NFV

Network functions virtualization (NFV) is fast becoming a classic "point of disruption" in the network technology market, and is emerging as a key pillar of the software-centric network. Operators want NFV to lower costs, to be able to scale services "elastically," and to benefit from greater programmability and service agility.

There's one huge problem, however. A primary objective of NFV is to substantially reduce the money operators spend on network hardware, and then to divert an ever-larger portion of the spending that remains to IT vendors, preferably no-name server vendors from the Far East.

You can see how that would be a hard sell to the vendor community. For some product lines, and some companies, NFV is perhaps even an existential threat.

One executive at a major vendor I met with at MWC explained the dilemma in down-to-earth terms: "How do I motivate a sales team to sell a version of the product that has 30 percent less revenue attached?"

A 30 percent revenue cut is also charitable. In this case, the product in question was an IMS application for the 4G core, which is sold on a licensing model, so 30 percent might be in the right ballpark. But we also hear operators say they're looking for much more substantial savings -- up to 90 percent in one case.

For smaller, software-centric vendors, this is not a problem. NFV is in fact a welcome development, because these firms have never had the capability to develop and test against hardware platforms the way larger vendors can.

But for these large vendors -- the market leaders -- what's the upside?

Consider also that applications will need to be at least partially rewritten to work optimally in the cloud. Spending on R&D to lose money on the eventual sale is hardly an alluring prospect. As a result, vendors are positioning themselves as ecosystem and platform builders, while quietly reminding operators of the value of purpose-designed, engineered systems.

Operators know all this, of course. They see vendors digging in their heels (not in public, of course) and proposing halfway-house approaches to NFV as more realistic and practical. The question is how this game plays out.

In our work with operators, we frequently hear statements like: "We will hold their feet to the fire"; and "We'll buy from startups, they can move at the pace we need"; and "Traditional vendors are being cut out."

The threat is that operators, working with IT vendors or other disruptive players, will seek to cut out "classic" network hardware and software products if incumbent vendors prove unwilling to move quickly enough to NFV.

Yet operators also know they need suppliers with long-term commitment to R&D and product support. And they know that without vendor buy-in, NFV will be much slower. The risk of moving without the major vendors is very high -- probably too high for most operators.

For their part, vendors see that operators are starting to put their money where their mouth is, and they are aware that they can't afford to find themselves on the wrong side of history. It is not that operators will spend less on the network -- they won't -- but they will direct investment to different places and functions. It will take vendors time, and judgement, to adjust to that.

Operators, rightly, want more networking capability for their money, and NFV can help deliver it.

— Gabriel Brown, Senior Analyst, Heavy Reading

brookseven 5/30/2013 | 1:17:57 AM
re: One Problem With NFV I think one needs to think about the quality requirements that the carriers have put in place as well as the vendor size requirements.

Imagine a 10 person SW company providing major infrastructure to a carrier...I think people need to take a deep breath and think this through. In my work in the SaaS space, we ran at a much higher ops cost because of the challenges of smaller vendors. A carrier might be trading Capex for Opex AND if the vendor goes out of business then the carrier is really stuck.

fstein 5/29/2013 | 11:57:26 PM
re: One Problem With NFV A few observations about lower price per box / per function.

1) Demand is doubling every year - at about Moore's law. So the systems should match and we're all revenue neutral.

2) NFV is more 'transformation' than 'disruption'. Supporting $100's B's (USD) of IB and physical assets, means evolution. Smart suppliers pick high ROI areas to virtualize.

3) The competition is not so much the other equipment vendor but the Web 2.0, OTT, Cloud, etc... If the 'function' goes outside the Carrier, so does the money.

4) NFV can accelerate innovation to build new services and new monetization models. The Carrier's fiduciary role can be leveraged for new services and partnerships.
afewell 4/24/2013 | 3:31:56 PM
re: One Problem With NFV Great post and very true points. One factor in play here is that there are large vendors that are already trusted SP suppliers that are not currently entrenched in L4-7 network services looking to move upmarket that dont have appliance revenue to lose. Whether they build nfv solutions organically or buy startups, there are capable disruptors afoot
joanengebretson 4/18/2013 | 6:12:52 PM
re: One Problem With NFV One of the drivers of NFV is to prepare carriers to operate networks that are required to carry higher traffic volumes without raising the price to end users. If networks keep growing at their current pace, we could get to where manufacturers make the same revenues on equipment that cost 30% less because carriers are buying 30% more of that equipment.
Gabriel Brown 4/18/2013 | 9:56:40 AM
re: One Problem With NFV I agree (I think) in the sense that identifying use-cases and picking off the applications to move forward with, will move the market. M2M, IMS, EPC, etc.

And it will require, as you say,-áthe big vendors to right-size their offers. This could be painful depending on timing, and so on.

There is also concern inside operators about "execution risk". Most of them know they are not sufficiently skilled, at scale, to pull this off across the network right now. Hence their focus on identifying feasible use-cases.

Thanks for the comment.
Connecter 4/18/2013 | 7:30:45 AM
re: One Problem With NFV Good article thanks Gabriel. I do not see a "huge" problem as you suggest. The elephant in the room is the ability for the "big vendors" to right size their offerings to the market. The operator market WILL move because it has to to meet the end user market - the emerging world of m2m and people/app mobility. Quality IT platforms such as IBM, HP, ORACLE will underpin the next gen NFV solutions because THEY will be the new "big vendors" that operators will look to for confidence and risk mitigation.
NFV (if you want any label) is here and real and in 3 years time it will be a very significant part of the operator landscape if for no other reason other than cost efficiency and scalability.
Sign In