I was recently reminded of a saying a former Juniper colleague used to quote about datacenter networking and relevant to network virtualization market development, which goes something along the lines of:

Every datacenter consists of four classes of network devices:  SANs, LANs, Firewalls, and Load Balancers.

His point was regardless of the specifics (i.e. size, applications, location, organization, in-house or outsourced operation, etc) about ANY datacenter you could ALWAYS find some instantiation of these four product categories in EVERY datacenter and hence understand where datacenter networking spend is concentrated.  Before anyone gets up in arms about the plethora of networking technologies omitted from this list – which is a topic for future blog post – perhaps we focus on these four technologies and what their presence in every datacenter means to network virtualization from a customer adoption and market development perspective.

Typically, customers adopt (or consume) technologies in one of two ways:

1. Point Products:  Where the customer purchases a point solution that solves a specific problem (i.e. one solution for SAN, another one for LANs, another for Firewalls, etc).  Generally speaking – point solutions tend to be the winners in early markets –  like NetScreen was for firewalls or as F5 is for load balancers.  These are bought to gain new capabilities and more often than not, in a large market, these products evolve to a very repeatable sales model because they have a limited problem scope to solve.
2. Platforms:  Where the customer purchases a platform that enables multiple sets of functionality from one technology.  Generally speaking platforms tend to win in mature markets where the main value created is in the integration of one vendors technologies – think Oracle or SAP – or aggregation of multiple vendors technology – think ArcSight or Splunk to integrate all of your log data.  Platforms are often bought to save money or do what you already do better or easier.  Compared to the Point Products, Platforms generally have  a relatively long sales cycle, as they require deeper customer commitment to integrate into their business and solve a ‘bigger’ problem.

If we think about a) the concentration of customer spend (i.e budget $’s) combined with b) how customers are likely to adopt a new technology (SDNs) – one could draw a number of conclusions:

1. At least as SDN relates to datacenter networking – the most financially interesting use cases are where budgets are allocated: SANs, LANs, Firewalls, and Load Balancers.  This doesn’t mean there are not other valuable use cases – it means that solving networking problems in these four areas are relevant to a) the most number of people, b) likely offer a path to a more repeatable sale; and c) for an investor – likely offers the potential for highest shareholder return.
2. Vendors that initially focus on building point products likely have a better chance at establishing early SDN revenue (and install base) to use to fund future evolution into a platform, than vendors who invest to build the platform up front without a broadly applicable use case that drives initial revenue.  The debate isn’t if there will be a market leading network virtualization platform — that will happen over time — the debate is the path to get there.  The risk with a Platform play is without customer demand, it’s hard to get people to write apps to your platform; without apps on the platform it’s hard to close early customers.

If I were CEO or a corporate executive in datacenter networking today and making a strategic decision to enter the network virtualization market, I’d be thinning about a) how we ship an initial point SDN solution around SANs, LANs, Firewalls, or Load Balancers in 6 – 9 months that b) leverages our existing products and sales capabilities, and c) provides flexibility to add a 2nd point solution once we’ve established leadership with the first solution.

Then, I’d evaluate the viable options available to get to market in 6 – 9 months; which given the early stage of SDN market development today likely include:

1. Build internally;
2. Externally fund a team to build to my spec (perhaps ‘spin-in’); or
3. Outsource development to a qualified software developer who understands network virtualization, SDNs and integration into existing networking technologies.
4. Explore partnerships for specific customer use cases that my sales team could use to close deals with today.

I keep outright M&A off the list for now — because, at this stage of the datacenter network virtualization market –  M&A is a hard sell to shareholders – there’s just too many unknowns, such as, customers are learning what they need; vendors are learning where they fit and most importantly what are their gaps to a viable product, and both customers and vendors are establishing the buying and selling process.  Implications to you are dependent on your point of view — for example, perhaps:

• Potential Customers:  Explore where SDN can add value to your business; consider funding a test bed.
• Established Vendors: Should focus and execute around what they can control.  Don’t let hype convince you buy an early stage company at this time.
• Startups:  It’s time raise that next round of capital for sales and marketing, focus on identifying a single, repeatable solution, your executive team should be leading the sales efforts to learn how to turn network virtualization into a repeatable sale.
• Investors:  Remember we are in the first inning, and you don’t have to rush to fund the first SDN company to knock on your door, nor do you need to pay a premium valuation.   Early deals are happening — and the key sales metrics to evaluate the health of a potential investment today should include: a) the ratio of professional services revenue to product revenue and b) number of prospects in the pipeline for a specific use case.

Like everything else – the views shared here are based on the best public information available at this point in time – and subject to change.  Perhaps, we’ll all learn a ton more at the ONS next week  that changes our view on network virtualization for the datacenter.

About the Author

.

Matthew Palmer

Matt has 20+ years of software-defined networking (SDN), cloud computing, SaaS, & computer networking… More

Connect with the AuthorWebsiteFacebookLinkedInTwitterGoogle+