[wp_tech_share]

Since 2017, no fewer than ten vendors have launched Smart Network Interface Cards or Smart NICs. The Smart NIC market is projected to become a $600 M market by 2024, or 23% of the total Ethernet adapter market. Vendors have developed or are developing innovative solutions to gain entry in the expanding Cloud data center market, and the emerging telco edge market.

Smart NICs, with an on-board processor, can provide a wide range of offload benefits in the following scenarios:

  • For a Public Cloud service provider operating a large-scale data center, Smart NICs could free up valuable CPU cores to run business applications for the end-user, potentially enabling higher server utilization.
  • Smart NICs are offered to meet a wide range of offloads. Some of these include transport and storage protocol offloads such as RoCE, TCP, NVMe-over-Fabrics.
  • Certain classes of Smart NICs are programmable and can be tailored for a wide range of applications and retooled to meet new requirements.

Smart NICs, however, are not without drawbacks and the following areas would need to be addressed before we see broader adoption:

  • Smart NICs are priced at a significant premium over that of a standard NIC. This price premium can be 5-10X higher given the same port speed and need to come down especially for volume production.
  • Smart NICs can draw anywhere from 20W up to 80W of power, which is not non-consequential on a per unit server basis.
  • Given the programmability and complexity of Smart NICs, they can consume significant engineering resources to develop and debug, resulting in a lengthy and costly implementation.

Given the above considerations the major Cloud service providers and IC vendors have developed Smart NIC adapters based on different IC solutions: ARM-based SoC, field programmable gate arrays (FPGAs), and custom ASICs. Each of these solutions offers varying degrees of offloads and programmability. In general, ARM-based SoCs and FPGAs are fabricated with programmable cores and can be adapted to a wide range of applications. However, the drawback of the programmability is the greater extent of engineering resources and lead time needed to bring the products to market. Custom ASICs tend to be hard-coded with customization generally limited to vendor-provided application tool sets. As products start to ramp and the market reaches consensus on product definition, we expect the following three categories of Ethernet adapters to emerge: 1) traditional or standard NICs, 2) non-programmable Smart NICs that are ASIC-based, and 3) programmable Smart NICs, that are ARM-based or FPGA based.

Network IC vendors that either currently have Smart NIC adapters or are planning to launch one have a wide range of solutions and target market segments. Notable vendors include Broadcom, Ethernity Networks, Intel, Marvell, Mellanox, Napatech, Netronome, Pensando, and Xilinx. The major Cloud service providers have developed their own solutions, further fragmenting market.

Our long-term outlook of the Smart NIC market by market segments is as follows:

  • Top 4 U.S. Cloud: In 2019, the Top 4 U.S. Cloud service providers, with Amazon in particular, drove more than 90% of the Smart NIC market by port shipments. This may be a challenging market for Ethernet adapter vendors to enter given that some of these Cloud service providers are likely to continue to develop their own solutions.
  • Other Cloud: These segments include Chinese Cloud service providers, such as Alibaba and Tencent, and Tier 2 Cloud service providers such as Apple and Oracle. As these companies scale data center capacity higher, Smart NICs could be a solution to enable higher utilization. These companies may not necessarily have the resources to develop their own Smart NICs, and are likely to seek 3rd party solutions from adapter vendors.
  • Telco Operators: This segment is increasingly looking to shift core network services to run on x86 servers, thus, Smart NICs could be used to offload network function virtualization. Certain adapter vendors are also targeting the emerging edge computing market as well, as Smart NICs is complementary to multi-access edge computing (MEC) nodes to satisfy low-latency requirements.
  • Enterprise: Generally, enterprise data centers tend to operate at a smaller scale, and would have less incentives to maximize utilization. Many enterprises would also rely on vendors to provide a solution with software implementation in place. Certain workloads, mainly enterprise storage arrays, are being developed with Smart NICs to facilitate NVMe-over-Fabrics connectivity.

As the Smart NIC markets continue to evolve, we believe the success of each vendor depends on whether its solution is a worthwhile upgrade over standard NICs from a performance, price, power consumption, and implementation perspective in their respective target markets. In 2020, activity level is high, as vendors work with end-users to complete product evaluation cycles. We expect to see volume ramp from a greater mix of vendors next year, as some of the short-comings mentioned above are addressed, realizing the benefits of Smart NICs.

[wp_tech_share]

With virtual MWC coming to an end, we are sharing some of the RAN related highlights below. To access the entire summary, please contact Daisy@delloro.com.

COVID-19 – It is no longer just about the supply chain

When we updated the 2020 outlook in conjunction with the 4Q19 RAN report, we assessed the potential supply chain risks as a result of COVID-19. At the time, the suppliers were not overly concerned about their abilities to meet 2020 shipments. While this assessment still holds, the overall situation has evolved since we published the report. In addition to any potential supply chain risks, other variables to consider include:

  1. The broader impact on the markets, the economy, and the increased risk of a recession – significant changes in GDP output could impact capex.
  2. Impact on cellular data traffic as people cut down on commuting and social events and spend more time at home.
  3. The suppliers might figure out how to minimize supply chain risks, but even with differing revenue recognition practices, sooner or later someone needs to deploy the equipment in the appropriate location.
  4. Forex exchange movements could affect the RAN market in U.S. Dollar (USD) terms if tightening financial conditions and aggressive stimulus by the Federal Reserve will move the USD relative to the Euro, SEK, and the Yuan.
  5. Governments will likely try various stimulus packages. Tax incentives to spur capex growth could be on the table.
Massive MIMO – Strong focus on performance, bandwidth, weight, and power consumption

The Massive MIMO business case has changed rather significantly over the past two to three years with the technology now considered to be a foundational building block for mid-band NR deployments. And with Massive MIMO configured 5G NR systems accelerating at a much faster pace than expected in 2019 and the overall installed base now in the millions, it was not perhaps a major surprise that the technology for a second year in a row remained one of the leading RAN related MWC topics (MWC 2019 Massive MIMO blog, see link).

But what was interesting this year was the different focus among some of the leading suppliers with Huawei and ZTE focusing more on their respective bandwidth advantages while Ericsson and Nokia emphasized the performance benefits with their Massive MIMO systems.

Dynamic Spectrum Sharing (DSS) to play a pivotal role in 2020

DSS was a hot topic during vMWC, validating the message we have communicated for some time, namely that the technology will play a pivotal role in upgrading existing low-band LTE sites to NR in the second half of 2020—Verizon recently confirmed it will launch DSS nationwide in 2020. The timing for any operator that is considering DSS will in addition to device availability to some degree depend on their overall 5G strategy – is DSS a 5G logo, an improved experience, or a stepping stone towards something bigger.

As the operators are doing their best to differentiate their overall 5G strategies, some confusion is inevitable. And with T-Mobile publicly suggesting that DSS is not ready for prime time and the technology its current state would impact the overall capacity negatively, the leading vendors took some time to clarify the state and benefits with DSS. But the key point here is that there are only three ways to move from 4G to 5G with existing LTE spectrum (re-farming, static sharing, and DSS). And if we are going to discuss relative performance of DSS, Ericsson rightly pointed out that we need to compare apples-to-apples.

5G is more power efficient than LTE with comparable cell consumption

LTE and 5GNR Power ConsumptionWith operators ramping up 5G while still maintaining their legacy 2G-4G networks, the power consumption KPI appears to be moving up the priority list as operators are trying to ensure total 2G-5G cell site power consumption remains within reasonable power budgets.

Since there is no shortage of media coverage suggesting 5G cell sites will be significantly more power hungry than 4G cell sites, the vendors have over the past year ramped up the engineering and marketing efforts with their 5G portfolios.

Huawei has on numerous occasions discussed the 25x to 50x (future 100x) power consumption per bit advantage between 5G NR and LTE while acknowledging that the overall NR cell consumption could be slightly higher. And now, partly due to continued channel, carrier, and symbol shutdown advancements, Huawei announced that the total power budget between NR and LTE is comparable when comparing the 64TxRx AAU with the 4×60 W LTE RRU.

Virtual and Open RAN gain momentum

There are multiple ongoing RAN virtualization efforts driven both by operators and suppliers with the primary objective of realizing a more flexible architecture with uniform hardware platforms that will optimize TCO and service differentiation for both the known and unknown use cases.

Given the current progress and the overall readiness with both the Open RAN and Non Open RAN virtualization tracks, we anticipate that the benefits with purpose built RAN will continue to outweigh the benefits with Virtual RAN over the near-term.

At the same time, the Open RAN momentum is moving in the right direction as the ecosystem develops, partnerships are forming, suppliers are ramping up investments, and operators are committing and experimenting with trials.

TIP announced a project targeting commercial multi-band radios by 2021 in the $1 K range. Even if the RAN is a relatively small portion of the overall site TCO, this cost structure is clearly compelling and would go a long way offsetting higher baseband compute costs typically associated with general purpose solutions.

For more details about the shift towards virtualized and Open RAN, please see separate blog.

For full blog access, please contact daisy@delloro.com.

[wp_tech_share]

Optical bandwidth is not limitless. I know we have all used the argument over the past few decades that by placing an optical fiber connection between two points you will have unlimited capacity. It is an argument we have all used when convincing the market that a fiber deployment future proofs the investment. Now, a few decades later, the reality of Shannon’s Limit hangs over service providers’ future like a dark cloud (an actual one, not the data center cloud kind).

Globally, service providers have always enjoyed the benefit of meeting customer requirements for higher network capacity at a lower per bit price by installing the latest optical DWDM transponders. It has been a tried-and-true method that leveraged the sunk costs related to fiber and amplifiers, often referred to as the fiber plant. Therefore, not only did a service provider benefit from the higher capacity and lower cost-per-bit inherent in each new generation of DWDM transponders, but they also benefited from installing these additional wavelengths in pre-existing fiber plants where much of the costs are either fixed (in the case of a leased fiber), depreciated (in the case of fiber ownership), or non-existent (in the case of fully depreciated fiber plants). The advantage of this is that for every bit added to a fiber, the cost of transporting the bits declines.

Even in the case of a green field deployment, the more a service provider can maximize the capacity of the fiber plant, the lower the cost of each bit transported. For example, if a fiber plant cost $4 million, the first gigabit may cost $4 million, but when 9,600 Gbps is installed, the cost-per-gigabit drops to $417 (Figure 1). It is a linear decline in cost. That is… until the capacity on a fiber can no longer increase due to Shannon’s Limit and spectrum availability.

Cost-per-Gigabit for $4 Million Fiber Plant
This constraint, brought about by the physical limitations of glass and light, brings to the forefront two challenges—extending the span length of higher wavelength speeds and increasing fiber capacity—and, fortunately for the industry, two solutions that will help service providers continue to reduce their cost-per-bit for at least another decade.

  1. 800 Gbps-capable transponders with probabilistic constellation shaping operating at around 96 Gbaud solves one of the challenges—extending the span length. Using these new transponders will help in a couple ways. For metro spans, service providers will immediately benefit from having a lower cost-per-bit by deploying one set of 800 Gbps line cards rather than two sets of 400 Gbps line cards. The second benefit is the extended reach. What was not shown in Figure 1 was that the un-regenerated span length of each wavelength speed diminishes as the wavelength speed goes up. However, with 800 Gbps-capable line cards, a 400 Gbps wavelength will be usable in many more fiber routes that extend in to the thousands of kilometers, reducing the need for costly regeneration sites. As a result, in both metro and long haul routes, 800 Gbps-capable line cards will reduce a service providers’ cost-per-bit.
  2. Expanding the usable spectrum in a fiber will solve the second challenge—increasing fiber capacity. For the longest time, optical equipment was designed to operate in 4 THz (80 channels @ 50 GHz) of fiber spectrum located in the C-band (Figure 2). Over time, equipment manufacturers increased it to 4.8 THz (96 channels @ 50 GHz). The next generation of equipment will inevitably be designed to operate in 6.0 THz (120 channels @ 50 GHz) of spectrum, referred to as Super C-band by Huawei. This action alone increases the amount of bandwidth-per-fiber by 25 percent. As a result, a fiber that had a maximum capacity of 38.4 Tbps (at 400 Gbps per wavelength) can now support 48.0 Tbps. Using our $4 million example in Figure 1, the cost-per-gigabit will be further reduced from $104 to $78.

Huawei Spectrum in C-Band

We believe these new technologies will need to be adopted in the future to continue the cost-per-gigabit declines that the market has gotten accustomed to.

Dell'Oro DWDM Long Haul Cost per GigabitAs a point of reference, in the past decade, the average annual decline in cost-per-gigabit for DWDM Long Haul transport equipment was about 20 percent due to improving spectral efficiency gains brought on by deployment of newer, higher-speed wavelengths (Figure 3). It was a similar rate of decline for WDM Metro, as well.

It probably does not need to be said, but this decline in the cost of a gigabit was a critical element needed for service providers to profitably keep up with network demand, which grew at an average annual rate of approximately 35 percent. Also, as we see it, demand for bandwidth will only grow from here, placing additional strain on service providers to profitably increase their network capacity for many more years.

We believe that with the new technologies mentioned, this pace of declining costs can continue to benefit service providers for at least another decade.

In the face of Shannon’s Limit, installing systems that can extract higher value from installed fiber becomes critical. Two technologies entering the market will push out the concerns brought by Shannon’s Limit for at least another decade. Therefore, to all the service providers, take a moment, and breathe a sigh of relief because there are new solutions on the way to help you: 1) increase network capacity and 2) reduce your future cost of bandwidth.

[wp_tech_share]

DellOro Scalable Infra CapexLet’s face it, cable equipment vendors are certainly happy to put 2019 behind them. A glut of DOCSIS channel capacity, the lack of significant competitive threats, and indecision around DAA technologies and timing all resulted in a spending slowdown that lopped off 35% of total DOCSIS infrastructure revenue, year-over-year. Traditional, centralized CCAP platforms bore the brunt of the reductions, with total revenue down 41% year-over-year.

On the bright side of things, each successive quarter in 2019 showed an improvement in scalable infrastructure spending by two of the world’s largest cable operators: Comcast and Charter Communications. In fact, the fourth quarter of 2019 saw a return to a normalized spending level for the combined operators, with Charter having bumped up its spending to increase DOCSIS capacity across strategic areas of its footprint. 2020 should see an improvement in spending by these operators, though again not to the levels seen in 2017 and 2018.

Also on a positive note, spending on Remote PHY equipment and Virtual CCAP platforms were both up solidly for the year, as a growing number of operators began their long-term transition to distributed and virtualized architectures. That trend will only continue to ramp up over the next few years, as operators continue to modernize their networks to push fiber deeper, reduce MERs (Modulation Error Rates,) and reduce the overall costs of operating their broadband access and outside plant networks.

Focus in 2020 Shifts to Upstream Bandwidth

DAA and virtualization are but parts of major network transformation projects many cable operators are beginning or are expected to begin this year. With the DOCSIS 4.0 specification establishing a clear path forward, giving cable operators the flexibility to pursue either Extended Spectrum DOCSIS (ESD) or Full Duplex DOCSIS (FDX), operators can move ahead with their remote PHY and remote MACPHY deployments to solve immediate head-end and power consumption issues.

Also near the top of many operators’ strategic initiatives for 2020 is the resolution of one of the known liabilities of cable broadband networks: limited upstream, or return path, bandwidth. Cable operators recognize that one of the liabilities they have with DOCSIS is its asymmetric design. For the most part, competitors haven’t exploited this liability. But with FTTH being more widely-deployed, they can certainly point at cable’s lack of upstream bandwidth and how that potentially disrupts latency-sensitive applications, such as online gaming, VR, as well as the simple uploading of photos and videos to social media. With a growing number of telcos and ISPs now offering symmetric 1Gbps services, cable operators are facing increasing pressure to expand their upstream capacity.

Most operators are still providing return band of 5-42 MHz. Using a mid-split design can push the upper limits of that band to 85 MHz, with a high split design giving operators up to 204 MHz to work with in the return path. Some large operators have already started or completed their transition to mid-split, while others are jumping directly to a high-split architecture. At 204 MHz, cable operators can offer 1 Gbps of upstream bandwidth, which matches what many telcos are offering through their FTTH networks.

But these upstream upgrades also require significant changes to the outside plant, including amplifiers and taps. For many cable operators, this isn’t necessarily a bad thing, as these critical outside plant components are nearing the end of their lifespan over the next five years, after having served in broadband networks for the last 10-20 years.

vCCAP platforms can serve a key function during this transition, by giving operators more flexibility in the approach they take and which part of the outside is impacted in certain upgrade cycles. When new capacity is required, both downstream and upstream, new vCCAP servers can be quickly added in any location or existing software resources can be re-allocated to those service groups undergoing capacity upgrades. In many cases, those resources can be added much faster than they can be with a centralized CCAP platform, which would require a linecard upgrade at a minimum to support increased capacity.

Though these upgrades put a lot on the plates of cable operators worldwide, the combination of all these transitions will ultimately lead to the complete overhaul and modernization of their broadband access networks, as they continue to migrate down the path towards 1.8 GHz spectrum, Extended Spectrum or Full Duplex DOCSIS modulation, DAA, and virtualization. Again, all these changes will definitely occur in phases, as each operator weighs the vision of their future networks and services against the short- and long-term costs to get there.

[wp_tech_share]

Every year vendors rally around certain words and the RSA Conference 2020 was no different. It had been about ten years since my last RSA conference after regularly attending in the 2000s. After nearly 40 hours of meetings and vendor conversations at Moscone Center in San Francisco, I kept hearing how vendors were inventing (or reinventing) themselves to deliver SASE, SD-WAN, Zero Trust Networks, and/or SOAR with a sprinkle, or even a good helping, of SaaS.  Let’s break down what’s behind the buzz:

  1. SD-WAN (Software Defined – Wide Area Network):

On the first day, the IT gods gave us routers for each branch office and expensive dedicated links. Eventually, businesses got tired of paying the high price for dedicated links. Plus, they wanted to improve the user/app experience that at times could get unstable, full of lag and generally poor. Behold: the SD-WAN routing solution was developed that can make snazzy forwarding decisions based on rich set of inputs, including user, app, latency, congestion and more, to deliver better user/app experience over cheaper commodity Internet links. It turns out many of these SD-WAN solutions also aim to provide network security. But many are not as effective as pure-play security solutions, such as FW/NGFWs.  Just so happens that the barrier to entry to implement SD-WAN isn’t like doing carrier routing. Thus, several security vendors have added sufficient routing capability to claim SD-WAN functionality.  At RSA 2020, I had some good discussions with Fortinet and Palo Alto Networks, both of which are investing heavily in the space.

  1. Zero Trust Network:

Back in early 2000s I helped get Network Access Control (NAC), an ancestor to Zero Trust Networks, off the ground. I ran the IETF RADIUS extensions working group, which developed some of the open authentication standards leveraged by NAC and, now, Zero Trust Networks.  While NAC flamed out for being ahead of its time, I found Zero Trust Network solutions at RSA conference as a more usable, superset form of NAC solutions from my yesteryear.  Zero Trust Network solutions are all about implementing a multi-segmented network by orchestrating between endpoints, access edge (campus, branch, cloud user edge), and applications/data being accessed.  At RSA 2020, the folks at Cisco and Juniper Networks walked me through their campus network solutions.

  1. SaaS (Security-as-a-Service):

While most think SaaS equals Software-as-a-Service, I prefer to think of it as an acronym for Security-as-a-Service. Boxes/appliances will never completely disappear, but the clear trend is to deliver services as a service where you want it and how you want it. Whether it’s NGFW/FW, email security, ADC/WAFs, web gateways, IPS, DLP, or ATP you can get it in a SaaS model, whether it’s a virtual machine, hosted service, or true pay-as-you go offering. Every vendor at RSA, or least everyone that wanted to be hip and cool, had a SaaS play.  This is a very exciting space for security vendors that I plan on digging deeper here at Dell’Oro.

  1. SOAR (Security Orchestration, Automation, and Response):

Everyone at RSA 2020 wanted to SOAR in some way. But to my mind, SOAR boiled down to the cool new name for a central dashboard for policy, visibility, and analytics. Fifteen years ago, we called SOAR’S predecessor the “Single Pane of Glass.” But with a decade and a half of refinements, SOAR seems to be soaring higher than the pain caused by the early-gen single panes of glass. At RSA 2020, I noticed the focus on bringing together a vendor’s product/solution portfolio with complimentary, third-party solutions. It looked like vendors had finally internalized the maxim that security can’t and shouldn’t always be delivered by a single vendor. Like SaaS, every hip vendor had a SOAR-type offering, whether or not it was referred to as SOAR.

  1. SASE (Secure Access Service Edge):

I saved the SASE (pronounced “sassy”) kid for last. As the new kid on the block, SASE is in that awkward phase that all young buzzwords/markets go through when industry lacks black-white clarity on what’s in it and who’s delivering it. This reminds me of the days when early cloud providers were hammering out the technicalities of IaaS, PaaS, and SaaS.

From what I can tell, we’ll continue in the storming and form phase a bit longer. But through the dust I can see that SASE, at its core, is anchored by an in-the-cloud gateway service. Thus, at minimum, SASE will replace on-prem security web gateway appliances with a side helping of Cloud Access Security Broker, Data Loss Prevention, applied threat intelligence, and even FW/NGFW capabilities brought together under a single administrative panel. At RSA, I had the pleasure of talking with the zScaler, Cisco, Palo Alto Networks, and ForcePoint SASE teams.

In future blogs, I’ll dig deeper into why I think these ideas caught wind. But for now, keep an eye out for a year filled with SASE security clouds over SD-WAN and Zero Trust Networks with some SaaS and SOAR… at least until RSA 2021. If you attended RSA 2020 and have different takeaways, drop me a note at mauricio@delloro.com. I always appreciate other perspectives.