A summary of what the top 2018 media stories imply about networking, including automation, multicloud, SD-WAN, NFV, 5G, 400GE, silicon-photonics, composable infrastructure, white box, unplugged consumers, and Huawei. We add some color to some of the most talked about topics in networking during 2018. In future articles we will cover our own thoughts on the most important issues.
Perhaps the hottest and most relevant conversation occurring at this time. Even if white box becomes pervasive, and it is far from clear it will, how much more CAPEX can be realistically squeezed, is questionable. This leaves operations cost as the next big rock to attack. Huawei made news in 2018 by saying network operations jobs can be cut by 90%. Other network vendors are emphasizing the opportunity to up skill workers so they can transition to higher value work, and new operations paradigms. Over the horizon of today’s workflow-oriented conversations is the aspiration of autonomy. Automation and autonomy promise to be important conversations in the coming years.
Software and Services
Multicloud heats up
2018 saw a much greater level of intensity in the multicloud conversation. Not just in terms of vendors talking about their solutions, but also in terms of vendor / cloud provider agreements. Google also announced go-cloud aimed at creating portable cloud applications. This is the area where we suspect the industry should be paying a great deal of attention, combined with various other container trends. If a significant amount of compute remains in private cloud for some time, as most believe it will, then the real action might be facilitating enterprises to easily choose which public clouds they can move workloads to. There are obviously networking and security components to this, but there also has to be the ability to run workloads on different clouds. There may even be a business model opportunity for someone to provide a public cloud brokering service. It goes without saying that one of the engaging aspects of this evolution is the competition between the cloud providers themselves, including some who have built-in enterprise customer bases that they are looking to leverage. It would be remiss to neglect the huge IBM acquisition of Redhat, Redhat having previously announced they were targeting VMware, and now having access to IBM’s customer/sales reach.
Multicloud is one of those trends that is hard to pick the timing of precisely in terms of huge revenue streams, because it depends on Enterprise transitions etc. But it does seem like a necessary direction, and there was definitely much more chatter about it in 2018.
Over the last couple of years, Cisco has acquired Viptela, VMware VeloCloud, and in 2018, even Oracle got in the act with the acquisition of Talari. SD-WAN, SDN, Multicloud, NFV, and managed services are all topics that have intersection points in current industry discourse. Those looking for big inflections in networking, often point to SD-WAN, claimed by some to bring all those intersection points together. We believe that MPLS VPN services are well-entrenched and trusted by service providers and enterprises, so transition to SD-WAN will not be over night. Promised cost reductions does create pull, even if only for extensions to existing networks. It might be that the company/ecosystem that seamlessly integrates datacenter and WAN overlay networks, and also demonstrates the value of doing so, will be the company that helps accelerate the move to SD-WAN / a common approach across datacenter, WAN, and whatever endpoints overlay segments need to be extended to.
NFV hits the bottom of the hype curve
That virtualized software should be a part of networking architectures seems intuitive. What the mission is, and what the technology framework for operations is, are both matters of debate. The early focus on replacing high-performance, highly reliable connectivity infrastructure is giving way to an openness about new directions, notwithstanding the existing interest in vCPE/uCPE services. Should the telecom ecosystem pursue a telecom-specific framework? We think the long-term trend is for telecom to move all IT away from telecom-specific frameworks, and we suspect NFV will ultimately be similar, even if shaped by some telecom-specific data models. The increased use of the term “cloud native” may suggest this shift is underway. 2019 will be an interesting year to see whether NFV claws back from the bottom of the hype curve (the “trough of disillusionment”) or if it morphs, via cloud native, to something different.
5G starts to look real
While 2018 was hardly the “year of 5G”, it was certainly the year where 5G started to look real. While deployments remain small, the journey has begun. Everyone in networking should be well down the path of having a strategy that accounts for the impact of 5G. As 5G is touted as being an inflection point for new architectures, including NFV-based mobile networks, 5G could easily be in the software section of this summary.
2018 saw announcements from Arista, Cisco, and Juniper on 400GE. Dell’Oro was quoted by media as expecting only a small ramp in 2019, with ports tripling in 2020. A big issue with 400GE is the availability of optics.
Cisco Acquires Silicon-Photonics Luxtera
With Juniper Networks having already acquired a silicon-photonics company, optics component dynamics, and the need for speed in the Data Center, Cisco’s acquisition of silicon-photonics company Luxtera is noteworthy, especially if it is true that Cisco beat out Intel and Broadcom for this deal. The generally larger industry focus on software & services, makes a narrative around optical layer differentiation, economics, and supply something that sticks out on the hardware side of the business
Touted as the next phase beyond hyperconverged infrastructure, by delivering agile / scale out pools of networking, storage and compute (perhaps memory as well?), Composable infrastructure as both hardware and software layers, and hence and interesting bucket of competitors including, at least, HPE, Dell, Nutanix, Netapp, and Juniper Networks (HTBASE) [Note, Juniper Networks also has a partnership with Nutanix]. While I put this topic in the hardware section, it is fair to say, significant innovation in this area is based on software.
White box aspirations remain high with some large purchasers. The reality is mixed, if not underwhelming. Beyond the well known successes in segments of cloud provider networks, 2018 included press stories about ATT’s plans for 60,000 white box-based cell-tower routers and a moment of false alarm around Amazon’s white box plans. The desire for white box is clear among large network / cloud providers. We expect ongoing evolution in this direction, but with the need for vendor-supplied, scale-up, high-performance routers for some time. This will be an ongoing conversation in the industry
We are heading towards a milestone in the evolution of the television experience. While unplugging has been emerging for over a decade, 2018 saw the first hints of an accelerating move towards live TV been easily accessible from streaming services, whether a telco offering such as AT&T’s Direct TV Now, or Roku. Once all, or even most, of the high-interest live TV can be streamed, it is hard to imagine why being unplugged would not be the default.
In 2018, Huawei continued to work around security concerns, including an endorsement from BT’s Neil McRae that Huawei is far ahead of the competition in the area of 5G. That said, the new year has been busy for Huawei including employees arrested in Poland and Canada. When elephants dance, everyone else has to be careful they do not get stomped on. Where the competition between the USA and China goes in 2019, and what the fallout is, will surely be one of the big themes in 2019. We may know whether USA and China can come to an amicable trade agreement before the middle of the year. If there is much at stake for all networking players in the transition to 5G, then there is much at stake when it comes to Huawei’s fate in both legal courts, and the court of public opinion.