I do have some experience in the networking field, and friends in the industry who I've spoken to about the subject, and I'd like to clear up some myths about the networking industry, especially with regards to this new super growth networking companies.
The traditional explanation for this "disruption" has to do with software defined networking (SDN). A very short version of this very technical subject is this - SDN transfers control of the network topology, comprising the configurations of many network devices, to a centralised controller instead of the many devices in a network needing to be configured individually (and often physically). This allows "on demand" allocation of network resources to the numerous incoming and outgoing packets, compared to the traditional model where each set of network devices was configured for a certain network LAN and each LAN had to have a large bandwidth buffer (i.e. a bandwidth capacity with a margin of safety to the maximum expected bandwidth). The latter was inefficient, with utilisation rates being estimated as low as 15%. But by combining the network resources together, the expected bandwidth spikes fall greatly relative to the total bandwidth used, so companies needed to install less bandwidth capacity overall. This is a very short summary of the effects which does not include the subtleties and benefits of SDN - it skips a lot of stuff about more flexible network management, higher security and visibility et cetera.
@kelvesy and @Barefoot
What most people jumping into this hot industry don't realise is that the biggest SDN company by market share is actually
Cisco. Cisco does in fact produce very good and popular SDN products. Cisco's revenues in routing and switching have not been declining, they have just been growing very slowly, in low single digits. Why has it been growing so slowly, and Arista, Juniper et cetera so quickly? The answer is a mix of a number of factors. The first and most easily quantified is that the global networking industry is growing only at a moderate single digit rate, about 4-8% and Cisco holds an enormous market share, about 50% as of 2018. So there is a limit on its growth. The second is that as mentioned above, SDN does lead to some reduction in the needed bandwidth. In the industry the pricing is price-per-port (for a port with a certain bandwidth, i.e. 100GB). So while Cisco does grab a lot of SDN market share, it does so at the expense of its traditional products, cannibalizing its own revenue. The third is traditionally explained as "price", which is only half of the story. It doesn't explain how Cisco grew so much when all the while its products were so expensive. The other half is the much more difficult to quantify "quality". Networks, like utilities, is an industry with an enormous emphasis on delivering the service with zero downtime, which combined with massive spikes in demand, means the backend needs to be well engineered, capable and robust. This is an entirely qualitative opinion, but recently companies like Arista have begun producing products of comparable quality to Cisco's. I think it is much better said that new companies have begun offering products of comparable service, at a lower price. There was also the alternative explanation that Cisco's iOS, a unified command line on all Cisco products, had a network effect that gave Cisco products seamless integration and familiarity. But with the advent of interoperability between different brands of network devices and open source standards, this eroded.
What is also not very well known is that SDN is not a set of products, it is an architecture. Generally speaking it is composed of two parts - one, a group of network devices capable of being remotely configured and two, a centralised controller (a piece of software). All the major network companies sell both, with a varying degree of interoperatability and functionality between each other. This is a very large blow to investors dealing with SDN because while the controller is a piece of software very obviously built for SDN, there is no definition on what is SDN hardware and what is not. The broadest definition would be any remotely configurable network device, in which case Cisco devices supporting SSH and SNMP (two kinds of protocols for remote configuration), are SDN devices and Cisco has been selling them since the early to mid 2000s. The strictest definition would be only devices that support OpenFlow (an open source SDN protocol) and some other proprietary SDN protocols are SDN devices. The latter is also a stretch because its possible to install OpenFlow support onto older Cisco devices, but I think it is more useful because OpenFlow supports a much more seamless and robust SDN architecture. This reinforces my argument, which is that SDN is not a key factor in the rise of these companies.
@Wildreamz Arista is not a software company. Just 12% of its business comes from Services. High profit margins is a trait in the industry because of the technological barriers needed for entry, and companies have a much stronger preference for quality than price for the reasons mentioned above. A networking software play would be VMWare, one of its segments provides an SDN controller and according to
this it is the second largest SDN player, behind Cisco. Comparing Cisco's revenue and Arista's on a one to one basis is not fair as Cisco produces a lot more than routers and switches (which is what Arista solely makes in its non services revenue). Less than 50% of Cisco's revenue comes from routers and switches. Their services revenue is also misleading as Cisco provides a much wider variety of software services than Arista.
Predicting the future growth prospects of Arista is beyond me, and I think no one can do it reliably. But I did find this useful graph that shows the decline of Cisco relative to Arista. Note that high speed data center switching is just a part of the networking industry, albeit a big and growing one. If we assume this reflects the entire business of Arista, which is not completely unfounded, it can grow about 4 times before its market share reaches Cisco, after which it will grow at the rate of the global networking market.
![[Image: 18374402_14790337262767_rId5.png]](https://static.seekingalpha.com/uploads/2016/11/18374402_14790337262767_rId5.png)
Homepage
PR Newswire
![[Image: lATiStl.png]](https://i.imgur.com/lATiStl.png)