Is Network Function Virtualization (NFV) a market inflection point, or just an over hyped technology in search of a use case?
- NFV is designed to use x86 hardware, which translates to improved capital efficiencies compared with dedicated hardware implementations.
- Software-based NFV deployment alongside real-time SDN network programing results in rapid service introduction and improved operational efficiencies.
- As NFV enables the decoupling of network functions and their physical location, services can be instantiated at the most cost effective location, in addition to multi-site application availability, scalability and cloud burst real time deployments.
- End to end service life cycle management resulting from common automation and operating procedures translates into operational cost reduction.
- Open and standardized interfaces between virtualized network functions and the infrastructure enables service provides to avoid vendor lock in. Each service provider can select its own best-in-class options and run in a multi-vendor environment.
Migration from legacy networking infrastructure design to a virtualized one reminds me of the transformation to what was called Softswitches in the early 2000’s.
In those days all voice switching was done by big-iron, custom-installed voice switches that took up an entire room and ridiculous amounts of air conditioning and power. The industry leaders were Lucent’s 5ESS and Nortel’s DMS100. The best part for these huge vendors was that these switches cost millions to buy and millions to install and maintain. Even more amazing is that at this time, the technology being shipped was more than 15 years old or more. This type of market share and technology lifecycle is unheard of today. In about 1997-1998 the technology, as a result of Moore’s law, enabled an entire room of equipment to be compressed into one third of a rack. If a more obvious business case exists, I haven’t seen one.
So why was no vendor yet successful at gaining share in the local exchange office?
The reason was the barrier to entry was huge.
Voice service at this time was still a carryover from pre-divestiture. Voice providers had tariff rules that meant they were required by law to deliver all the features in the tariff. Even defunct or minor features that less than 1% of the subscriber base used could not be turned off. This meant that a prospective switch manufacturer who wanted to build a new type of switch still had to develop the thousands of features that the Bell system had come up with in the previous 20 years.
Beyond features, the switches were required by law to be five 9’s reliable. This means they have to work 99.999% of the time. This comes out to about 5 minutes of downtime a year. Building this level of reliability in software is not easy.
The catalyst for change was the telecom act of 1996. Now start up telecom companies had a legal way to begin challenging the incumbent voice providers’ monopoly. These new providers were called CLEC’s (Competitive Local Exchange Carriers). Since the CLEC’s were capital constrained they were more interested in trying switches that were being built by startup Softswitch vendors like Sonus, Telica, and Unisphere.
So now giant vendors like Lucent, Nortel, and Alcatel had a problem. If you build a switch to replace your legacy switches, won’t you cannibalize your cash cow? If you sell a switch to a CLEC for $500K that fits in a third of a rack, how can you justify selling the same switch to AT&T for $5M? We know how this story ends. Today, these giant vendors no longer exist. Of course there was more to it than just voice switching, but it was part of it. I know I lived through it, barely. Disrupt or be disrupted.
This brings us to the present day. Computers for the most part have been virtualized, meaning servers are using Linux open source operating systems. Server storage and microprocessors have been commoditized. Openstack orchestration is here. This leaves networking, and security.
Large consumers of networking technology are finding various ways to reduce their reliance on custom hardware. Why? Custom hardware primarily based on ASIC’s provide superior performance, scalability and reliability. Sound familiar? The 5ESS was probably the world’s biggest, most reliable, best-performing, custom-built computer ever devised. Thousands of them still exist in central offices around the world.
Consumers of technology want to leverage Moore’s law to reduce costs. NFV reduces costs first because it relies on commodity (COTS) compute, NIC and storage. But more, it provides flexibility in how it’s deployed, purchased, maintained, and ultimately replaced. Carriers, enterprises, hosting companies, e-commerce companies, essentially all the biggest users of networking technology are getting smarter about how they buy. They realize that planning for how they migrate (i.e. get rid of the old) is almost as important as planning for buying the new. It’s managing total cost of ownership. TCO includes the OpEx for running it.
So what does this mean for today’s networking equipment suppliers? The successful companies will embrace NFV and transition their marketing and selling motions to adapt to it. This does not mean:
- Paying lip service to NFV while pricing software at the same as, or close to the hardware solution.
- Pretending to support NFV while failing to keep feature parity with appliances, leaving the software version constantly playing catch up.
- Selling software like hardware with the same price structure and maintenance prices.
- Selling a software version of the appliance but capping its scalability to “protect the big machines.”
- Keeping the software versions buried deep on the price list and telling the sales force it’s there for you to put in labs or do POC’s, “Just in case you need it.”
- Paying the sales force increased commissions or incentives to continue to push the high margin “boxes.”
So what is the best way to migrate and evolve product, marketing and sales to NFV?
Create a software desk.
Software prices are not competitive. The result is all or almost all quotes and proposals may require executive level discount approval. Large discount levels on direct carrier deals are not uncommon.
The result of such a system is that budgetary quotes off the price list going to prospects may get you eliminated before a deal matures enough for the sales teams to get feedback that the price is too high.
Sales teams may not want to engage senior leadership to make proposals that aren’t fully mature. Creative non-traditional, rip or replacement-type proposals never get made.
Most carriers are requesting software-based pricing for comparison purposes even if they know they are going to buy appliances now.
Selling software solutions requires a more creative deal structure than creating quotes from a price list. It also allows for disruption and market penetration, and subsequent growth in market share.
- Perpetual vs term
- Centralized vs distributed architecture
- Innovative partnering requiring less deep integration
- Rev share models
- On-demand and burst mode distribution and activation
- Unique requirements with license transfers.
Sales leaders and representatives operate in the domain they understand. Software products are positioned reluctantly and then only when specifically requested by customers, then positioned and priced as alternatives in direct comparison to its purpose-built appliances.
Sales teams look at software as a messy alternative to only be dealt with when absolutely necessary. Concerns exist because:
- Configurations and recommended hardware requirements are unclear and not properly communicated and maintained.
- Revenue for software products are perceived to be lower, because of course you are selling less content.
- Fear of the unknown.
- Pricing has been non-competitive.
- Poor to non-existent tools and portals for license generation and tracking, resulting in poor customer experience.
Do not try and re-train or modify field behavior. Transition to software solutions will be gradual and strategies and sales points of emphasis will most likely iterate several times before market adoption becomes wide.
Pivoting a global, large sales force is difficult, expensive and risky.
The alternative is to create a resource that can support software sales.
Software desk can be brought in to support opportunities.
Software desk will have authority to broker unique deals without a lengthy complex approval process. Software desk frees up executive leaders from having to get involved in every proposal.
Software desk allows engagement earlier in the process so that deals can be flushed out before they become a crisis or fire drill.
Most importantly, creative offers can be proactively presented to prospects to drive competitive advantage. Radware can drive new use cases and flank the competition, breaking into accounts currently dominated by competitors.
Software desk is not just an approval function. More importantly it’s a resource to help the sales force proactively approach customers early in the sales cycle or even create opportunity where none previously existed.
Software desk can be a central repository for information around how and where the market is moving. Visibility and trending for competitors are doing. Information will allow PLM and senior management to better align products in the rapidly changing market.
How could it work?
Software desk acts as a resource and advisory role.
Sales teams still own the accounts and decide strategy.
Software desk will have authority to create and approve non-standard deal structure and pricing.
Software desk will be the first stop and then a fast track to executive approvals for sales teams.