In my last Musing I forecasted the emergence of a new device type called the Web Edge Processor (WEP).   I said that it is likely WEPs will be deployed in front of every server or collection of servers providing networked applications, basically turning the situation today, like Figure 1 below:

into the slightly more simple Figure 2:

Of course, nobody builds web application delivery sites like Figure 1 - because they can’t. It is impossible to get packets to flow through each device in the order required to implement the system requirements; you can’t get each product’s failover mechanism to work with each others to create a fully redundant system.  It’s too expensive, and it’s just too damn hard to manage. 

So, if end users are going to benefit from all of the value-added features and functions that have been and are being developed to augment web application delivery, some vendor needs to step up to provide a pragmatic way to deploy such capability.  It doesn’t necessarily mean it all has to be delivered from a single product platform, but at least from a single vendor via a rationalized product line architecture.   

Who is going to step up to this challenge?  Will it be the market leader? 

Before we answer that question, maybe we should step back and study who the market leader is.

It’s not Cisco.  F5 Networks has been the vendor at the forefront of just about every innovation and important shift in the market since its very inception.  They earned about $109M in the last year, which I estimate represents about a 20% market share, give or take a few points.  Cisco may have earned just as much or more, but it’s hard to pull out their revenues in this segment from their overall revenues.  Even if they earned more, however, Cisco does not influence this market as much as F5 does.  For F5, it’s all they do.  For Cisco, they can probably generate more margin dollars by negotiating with their suppliers to reduce the price of DRAM by 2% or by lobbying congress to not require the expensing of employee stock options than by developing and supporting a new traffic management feature, whereas F5 cares desperately about both growing the market and dominating it.  As a result, F5 is the leader, trendsetter, and innovator in this space.  Consistent with this designation, F5 Networks is the furthest along in terms of trying to help customers sort through the cacophony of web application infrastructure design by promoting multi-vendor solutions for via partnering, cross-product and application testing in their Solutions Center (announced earlier this year), and coordinated field support for multi-vendor solutions.

Will F5 dominate this space like Cisco dominates Layer 2/3 networking?  Maybe.  Cisco and F5 have some similarities. In fact, a few years ago I wrote that F5 was “the Cisco of Internet Traffic Management.”  What I meant by that was their style and winning formula were similar.  Some of the factors that made Cisco great in their formative (roughly 1985 -1992) years are shared by F5:

• ·     Amazing caliber of employees.  Way back when, I was involved in an OEM relationship between Cisco and AT&T and I remember being amazed by the across-the-board caliber of the Cisco employees.  And this wasn’t restricted to the engineers and product managers.  I remember thinking that their lawyer and reception desk clerk seemed to know more about the technology and market dynamics than we AT&T engineers and product managers.  It was quite a humbling experience (which I have since gotten over). 

I haven’t had a lot of interaction with F5 in the last year or two but prior I did and I was always impressed with the across-the-board savvy of their employees and especially the business sense of their management team.  In contrast, a lot of their competitors always seemed to be just kind of winging it.

• ·     Software focus In their early years Cisco was basically a software company.  They pre-packaged their software onto hardware to make their products more stable, dependable, and deployable.  Back then, using Intel 1U boxes as a platform wasn’t really an option, so they built their own hardware.  But it wasn’t a focus.  The hardware was a kludge, but good enough.    By conserving the amount of energy they put into hardware, they could focus more on software. And focus they did – on routing protocols and a lot of enhanced features that made scalable packet-based wide area networks feasible.  Some competitors - long gone now - pointed out deficiencies in their hardware and built better boxes, but in doing so the gap between their software capability and Cisco’s just got greater and greater, until Cisco just became the de facto industry solution.  Later, as the market rewarded Cisco for their software capability, Cisco used some of their earnings to go back and improve their hardware, but their success is rooted in their early focus on software.

F5 is similarly software-centric.  In their formative years, almost their entire R&D went into software-based feature development.  To make installation easier for customers, they pre-packaged their software onto PC platforms, helping to foster the “network appliance” concept, which has since become increasingly popular.  It was only later on, when they experienced some competitive pressures from Alteon and Foundry, and as their increasingly complex features required more infrastructure support to perform well, they started doing a little more hardware development and integration to create more scalable platforms.  But, as with Cisco, the “get the feature lead and then go back and improve the performance for high end users” approach worked well. 

• ·     Sales driven Amazingly for a high tech company, Cisco has not generally been Engineering driven, but has been Sales driven.   Cisco has always been geared towards domination and the focal point of their crusade has been their direct sales force, whose mission is simple – get the deal. Their unflagging aggression towards that goal led to a characteristic that is symptomatic of almost all successful high tech companies – the competition loathes their sales force.

      The competition hates the F5 sales force as well. They must be doing something right.

So… Cisco and F5 Networks have a lot in common. But there is also one very important difference between the companies.

Cisco created a tremendous technological barrier to entry to others by inventing a proprietary routing protocol (IGRP) that enabled large and complex routed networks.  Anybody that wanted to build a network that spanned more than a few hops or which included a wide range of link speeds, needed to use Cisco for their solution.  More than anything else, this single invention led to the proliferation of packet networking we see today, creating the $25-30B routing/switching market that Cisco dominates. Had Cisco not had such a critical advantage – had the technology playing field been more level – the routing market would have been more fragmented with more players and unit pricing would have decreased due to competition.  Because of the factors mentioned above, I still think Cisco would have won the war, but the spoils of victory would not have been as lucrative.  Cisco probably wouldn’t have gained the same financial muscle early on which enabled them to invest more heavily in their technology and channel and marketing, enabling them to pull further and further away from the pack, which continued the virtuous cycle for them.  Ultimately this also led to having the financial ability to acquire lots of companies to get into new market spaces or to fill in gaps in their product line. 

While F5 has developed a huge steaming pile of useful software that implement lots of important features and solve lots of complex problems – they don’t have the equivalent unfair technology advantage that IGRP provided Cisco.  It’s just a lot of hard work for competitors to achieve similar capabilities, as opposed to abrogating patents and other legal land mines.   

Another thing that Cisco has that F5 doesn’t have is a rapidly growing and huge market. Routing and switching is a huge market largely because it is physical – every wire in a network needs to be driven by a switch or router port.  And there are 100s of millions of wires in the world that are part of networks.

F5, in contrast, plays in a market sandbox that is much smaller because it is virtual.  A few F5 boxes can interdict and influence traffic flows for 100s or thousands of servers and applications. Unfortunately, competition has driven unit pricing down well below what the value of the technology is.  The result is a modestly sized - $500-700M – market, with flat to single digit revenue growth.  In this situation, the market leader doesn’t get nearly the financial leverage that Cisco had.  F5 Networks has about $90M in the bank, and a market capitalization of about $470M. If you or I had that much cash in the bank and had that much net worth, it would mean we wouldn’t have to worry too much about how much we spent on Christmas presents.  But for a company that has recently been losing money each quarter, it means they have to pinch their pennies and buy Hyundais instead of BMWs. F5 is not going to be able to go on an acquisition spree for healthy, growing companies.  

All this means that in this segment, the gorilla, although lean and mean and to be respected, is vulnerable.  F5 is not necessarily going to be the dominant WEP vendor, although I think the game is theirs to lose.

Although I promised it for this Musing, in next month’s Musing I’ll get more specific about what I think it is going to take to be the dominant WEP vendor, especially from a start-up perspective  Stay tuned.