We attended the Wi-Fi NOW conference in Redwood City, CA this week and attended some interesting presentations. We are writing about our observations and notes from the Google, Quantenna, Mist Systems and Mojo Networks presentations.
Google Station presentation. "GOOGLE STATION: PUBLIC WI-FI TO CONNECT THE NEXT BILLION INTERNET USERS." Monica Garde and Erika Wool made an interesting presentation. The jist of the presentation, from our viewpoint, is that Google is partnering with service providers and enabling these service providers to monetize the Wi-Fi network through a revenue sharing system that is based primarily upon advertising. The company shared some statistics, which we have in the accompanying slide.
Quantenna presentation. James Chen, VP Product Line Management presented "GREAT INNOVATIONS PART ONE: MASSIVE MIMO & DUAL-BAND 802.11AX". Chen made the the case that 8x8 WiFi (that Quantenna calls Massive MIMO) outperforms 4x4 systems. For instance, in its tests, at 85 RSSI and through a wall, performance was 1.6x greater using 8x8 compared to 4x4. The company also made the case that Massive MIMO has greater throughput compared to non Massive MIMO, as well; the company has demonstrated >1 Gbps throughput in a typical home. The company showed that Massive MIMO alleviates the "Sticky Client" using a 1x1 Samsung Galaxy Tab Active2 device. The company did not talk about 802.11ax, unfortunately, other than to say that 8x8 is relevant for 802.11ax, as well.
Mojo Networks presentation. Mojo CEO Rick Wilmer made the point that simply enabling Cloud-managed Wi-Fi has been done already, implying this is cloud 1.0, and that this message is boring. The company explained that its cloud architecture is cloud 2.0 because it takes advantage of the capabilities in the cloud and enables - Cognitive Wi-Fi. Cognitive Wi-Fi, as far as Mojo is concerned, has to do with big data (store key client parameters and run ML algorithms) and smart edge APs. The company didn't go into deep science of ML/AI, but explained the ML workflow: 1-data collection, 2-training the classifier model, 3-trained model in action, 4-result.
Mojo explained that it has lots of data to perform Machine Learning on. It has 1/2M APs deployed. The company shared that using 1 week of data of a subset across only 4 verticals (enterprise, education, mfg, retail & hospitality): 237K clients, 31M associations, 400+ applications. Separately, the company said it has obtains 50M associations per week (in a press release). A significant amount of the data that is delivered to the cloud has been pre-processed in the Mojo APs; the APs cache 2 days of data. The point of these statistics, according to Mojo, is that it has more data than other Wi-Fi vendors to train its Machine Learning system on.
According to Mojo, using inference engine, automatically fixes everything possible. Wilmer says that this makes interacting with the User Interface less necessary because it takes care of problems automatically. Was Mojo serious or joking when it said, "the UI may disappear as we know it." Time will tell.
The company shared some other information that was interesting:
Mist Systems. Bob Friday, of Mist made a presentation on May 17, 2018. In addition to the content from his presentation, I interviewed other at Mist personnel at the show. The company claims it is focusing and having success in selling to large enterprises. We learned that Mist uses Broadcom WiFi chips and has a custom-designed Bluetooth antennae array (shown at the show). The company highlights its location services as a unique capability, and it draws upon its Bluetooth capabilities to deliver location. However, the company's main message is its AI capabilities; in some ways, it has become the poster-child for AI amongst startups in the networking industry. Mist's presentation at the show reiterated the same point - that it is an AI company.
Stepping back, Mist has been shipping commercially for a year now. In our observation and research, the company's efforts to take share from competitors has landed it on the map - over the past two quarters, its larger competitors have taken notice of Mist and see it competing at large enterprise accounts.
During the Q&A part of the presentation by Bob Friday, Mist CTO and founder was asked something that we found very interesting; the question was what kinds of algorithms does Mist use in its system, and do they all need to learn? The answer was to the effect that many different types of algorithms are used, linear optimization, decision tree analytics, neural networks, etc. Friday made the case that there are just certain things you just know about how a WiFi network will and should work, so why go an have a machine learn about it when you already know it. This begs the question - how necessary is AI in the first place, especially if the vendor and its IT workers or VARs have gobs of experience and can design and implement a Wi-Fi network right in the first place. Looking at the problem differently, what this means is that some vendors may have had different backgrounds than competitors and can design Wi-Fi systems that know how to work under a variety of working conditions. Friday was also asked another question - given that Mist is focusing so much on AI, does this mean that far fewer IT workers will become employed? Bob's answer was diplomatic, but probably true - he said that no, we'll need the same number of workers in the near-term, and that AI Wi-Fi will simply allow the same number of IT workers to make better decisions. Still, the question makes it clear - the audience is concerned about job loss as AI works its way into the IT industry.
Earlier this year, Aerohive issued a press release about its A3 software system. A3 is what we categorize as an Enhanced Network Access Control (ENAC) system; we calculate market share statistics on this market in our Security report series. The company is now getting ready to bring the product to market and is blitzing the media, so to speak. We were briefed and learned more about the product. To summarize, it checks the boxes necessary for us to include it in our ENAC report and we like that it has a common user interface to allow customers to perform device profiling, authentication / registration, compliance / remediation, device management, billing integration and network access control.
Our outlook for ENAC is positive and, today, three main vendors consist of the majority of share: Cisco (with ICE), Aruba (with Clearpass) and ForeScout. We learned that company has more aggressive pricing than these market leaders. If you look at why a product like A3 is important to Aerohive, recall that back when Aruba introduced Clearpass back in 2012/2013, it used it as a selling tool to get into its competitors' accounts (it also got high margin sales from Clearpass, too). Aerohive's A3 is positioned similarly - it operates with its competitors' equipment (including those from Cisco, Ruckus, Extreme and others). So, Aerohive has developed another means of selling to customers, by offering A3 to customers using non-Aerohive equipment.
Tomorrow, 650 Group's Alan Weckel will be a featured speaker on the NBASE-T hosted webinar, entitled "Growth of NBASE-T, Market Trends and Forecast." In this webinar, we will review Ethernet trends that related not only to Campus Switch ports, but also WLAN, computing and other devices. The NBASE-T ecosystem continues to expand, including most recently with broadband modem devices. We are excited about this market and hope you will attend.
There were two main announcements, a new relationship with Google Cloud Platform and a new flash device - the AFF A800. Also, in our interviews with NetApp, we learned about the future of Fibre Channel at the hyperscalers.
Google. Google Cloud Platform now integrates NetApp Cloud Volumes as a drop-down menu capability as part of the Google console. This allows enterprise customers, for instance, to use Cloud Volumes to manage their data on Google's cloud service while simultaneously managing their data on premise. This relationship with Google now rounds out the NetApp relationships with the main hyperscalers - it already has in place relationships with both Amazon (AWS) and Microsoft (Azure). NetApp Cloud Volumes on Google Cloud Platform is currently available as a "preview" capability (sign up at www.netapp.com/gcppreview) and is expected to go to commercial status by the end of 2018. Customers will pay Google for the use of NetApp Cloud Volumes.
AFF A800. New flash hardware available from NetApp, which besides having impressive density and low-latency capabilities supports NVMe-over-Fibre Channel. Of course, the product also supports 100 Gbps Ethernet. From a historical standpoint, it is interesting that NetApp, a company whose heritage was driven by storage over Ethernet, is touting Fibre Channel. But, that's what its customers are asking for in order to accelerate their on-premise workloads such as database (Oracle), ERP (SAP) and other mission-critical enterprise workloads. In our interviews with NetApp, we were told that Fibre Channel is growing faster than Ethernet - this makes sense given the company's foray in recent years to flash and low-latency workloads.
Fibre Channel at the hyperscalers? We asked about what is going on with the hyperscalers' architecture to adapt to AI/Deep Learning workloads. NetApp executives explained that AI workloads are different from traditional workloads; they are random, low latency workloads connecting to GPUs. This type of workload, we were told by NetApp, works very well when attached to Fibre Channel. From NetApp's perspective, if customers want to run AI workloads fastest, they would likely do so on-premise, using Fibre Channel. Yet, many customers run their workloads on hyperscalers, all of which use Internet Protocol and the underlying Ethernet infrastructure. We have always been skeptical that hyperscalers would adopt Fibre Channel. We believe the hyperscalers may work with vendors such as NetApp to develop additional software capabilities to address the changing workloads relating to AI/ML/GPU workloads in the future - on top of IP/Ethernet infrastructures.
Today Arista announced its entry into the campus market launching several products in the core and aggregation layer for campus switching. For years there has always been a blurry line on how one defines campus core, partially driven by the utility of the Cat 6500 installed base. But this has been changing as the Cat 6500 installed base gets refreshed with purpose built boxes.
For the most part, when an enterprise has two separate networks for the campus (user connectivity) and data center (server and compute access), campus core is counted in campus. When campus and data center are one network or the location is smaller the campus core is usually a portion of data center. This different deployment scenario has caused confusion in the exact size of the campus core market.
In our research we look at it in both use cases to better understand the too unique use cases as businesses truly look at it differently depending on their networking heritage and IT expertise.
What is happening now in the campus market is that campus switching is changing. Campus switching is transforming from a user connectivity role to an infrastructure role to support the change is how users connect and to support IoT which started with just cell phones and tablets, but is about to explode. With campus connectivity changing, so is the core and this is allowing enterprises to rethink their campus core.
Many customers will continue to see campus as a separate network, but many customers, especially when they are building a hybrid cloud data center are looking at merging the line between server compute and campus connectivity. We will see this architecture change at the same time the market moves towards MultiGig in the access layer and towards 25 Gbps and 100 Gbps.
Keynotes at the NFV World & Zero-Touch Congress in San Jose, California were very interesting today. We share our observations and view of the main themes from these interesting presentations by Nokia, NEC/Netcracker, Google, CenturyTel. The main theme of these presentations, we think, is this: NFV/SDN is now deeply in the deployment and commercial phase, where compared to 3-4 years ago, it was just a concept.
Nokia. The company announced that its Airframe server platform, which is an OCP based design, comes available with either embedded acceleration or pluggable acceleration. This comment includes its software acceleration. The company explained that its Reefshark chipset can be equipped on the Airframe server and can perform better than a non-accelerated server:
In explaining functions that an Airframe with Reefshark can perform, the company gave a good example: massive MIMO beamforming can be assisted by the machine learning capabilities.
NEC/Netcracker. Enrique Gracia presented several uses cases of the NEC/Netcracker customers that related to NFV/SDN. He explained that 16 customers have deployed one or more of these uses cases.
Full Stack OSS/BSS/MANO. A customer deployed this system in 12 weeks to launch a VNF. The system managed both physical and virtual devices.
Expand to a new territory using VNFs from home region. A customer now delivers services to a customer outside the home territory by deploying the software and service from the network location at the home location. In this particular case, NEC/Netcracker and its customer do revenue sharing and VNFs include SD-WAN, virtual firewall and others. The service provider is expected to expand its customer addressable base by 40%, mainly targeting small/medium businesses in this non-home region. This system uses MANO, OSS, BSS and the marketplace. The company says in this case, time to revenue is expected to take 50% less time to deploy new VNFs in the future.
uCPE (Universal Customer Premises Equipment) deployment instead of branded hardware. The company worked with a service provider company to enable uCPE to be deployed as an alternative to Cisco, Juniper and others' gear.
Google Cloud. Vijoy Pandey, who represented Google Cloud, presented on the topic of using AI/ML to reconfigure its data center system. The company's cloud data center architecture has been evolving continuously since it was first introduced. Currently, the company is using its own AI/ML system to learn from current network traffic patterns in order to design its future network architecture.
CenturyTel. The company has deployed Broadcom based Ethernet switches using its own Network OS. These switches do their own packet forwarding. Additionally, the company has built its own orchestration system called VICTOR. It draws upon Ansible, NetCONF, uses the service logic interpreter from ONAP and uses parts of Open Daylight. The company plans to open source this development and the spokesperson Adam Dunstan said, perhaps jokingly, that this might be called ONAP-lite.
We attended the Huawei Analyst Summit 2018 in Shenzhen, China last week and reported mainly on service provider trends. This week, we highlight some of our observations about Huawei's Enterprise business, specifically, the enterprise business.
The Enterprise Services revenue grew 33% Y/Y in 2017, driven by 56% Y/Y growth in China. The company said it does more enterprise services revenue outside of China, but that China is its largest region. Additionally, the revenue growth seems to be tracking with its overall growth in CSPs (it's services partners), which grew in number by 34% in 2017. The company said that it has historically worked with regional services partners, but it found that in engaging with larger multinational companies, it needed to expand its partner base to include companies that had a presence in all major regions of the world, so in 2017, it signed up with OVS, Arrow and an Asia Pacific company.
The Enterprise Services group plans to increase its subscription services, which today is mainly its SmartNOS offering, which was made available 1-2 years ago. In pursuing this plan, the company is developing automated systems. The company has recently been pursuing expertise in vertical industries and has been selling solutions, instead of just products, it says. In recent years, we have observed this solution selling strategy by Huawei, and it appears to be allowing the company to take share from more product-oriented sales strategies of other vendors in the Europe, Middle East & African geographic theatre.
By and large, the revenue in 2017 from Enterprise services is derived almost exclusively from maintenance revenue on hardware sales, plus SmartNOS, so the company's goals represent a significant pivot from the current way of doing business. This signals a significant change in the way Huawei's Enterprise business plans to approach the market, in our view.
The main theme of the the Huawei HAS 2018 meeting keynotes was Artificial Intelligence and, secondarily, nearly ubiquitous networks connections across the world. Huawei expects 86% of enterprises to have experimented with AI by 2025 (<5% in 2018). It is leveraging AI across nearly all its products and will offer a full stack AI solution to all Huawei partners at its @Huawei Connect 2018 conference (Oct 10, 2018).
More specifcially, Huawei is using AI to elevate products & solutions to new levels: cloud, networks, devices, EI, SoftCOM AI, and Intelligent phones. The company's strategy has changed over time and is now AI-focused:
2006-2011: Single strategy: All IP
2012-2017: SoftCOM: All Cloud
2017+: All Intelligence: SoftCom AI (autonomous networks / services 2.0) - this reduces operating and maintence costs
The company expects that networks will be 10x more efficient in the operation of equipment as a result of AI.
By 2025, Huawei expects 440M AR/VR users, 40% of cars to be 'connected," 80% of users with access to mobile broadband, usage of 1 Gbps / user / day (versus .03 in 2018) and 20B connected devices worldwide.
Connected Devices Forecast (Huawei) by 2025: 40B sensors and 100B connections. This thinking is based on data including that there are:
The company's product lines are very diverse; to wit, the company introduced a helmet for the blind, which will be available soon.
Huawei expects NB-IoT (LTE-based IoT capability) to reach almost full coverage in China in 2018. Additionally, the company expects NB-IoT to reach 100 networks by the end of 2018 (versus 39 in 2017) and to be available on 1.2M base stations (from 0.5M in 2017) and to be connected to 150M connections (versus 10M in 2017).
The company boasted about several developments:
Q&A after keynote:
Mr. Eric Xu, Rotating Deputy Chairman of Huawei dodged several important questions relating to trade tariffs, cloud business unit revenue targets, growth rates of each major business units, specifics about AI full-stack claims made during the keynote, and instead focused generally on the AI theme. Xu did, however, however, answer a handful of questions that were quite interesting: Huawei won't acquire DRAM, Flash companies; and that 5G is not so revolutionary - it is just an evolution following LTE. Additionally, Xu mentioned that in 2H18, Huawei will launch end to end 5G solutions and by 3Q19, it will launch 5G capable phones. Xu said Huawei will continue to work with Intel on x86 for the foreseeable future.
More Q&A specifics:
Trade Tariffs and ZTE. (In a moment of levity, however, Eric Xu smiled when the words ZTE were mentioned - recall that a day earlier, ZTE was penalized by the US). We will focus on our customers and will ultimately survive.
Cloud 1.5B by 2020, will you hit the target? Will offer cloud services to telco service providers. Huawei smartphones will leverage the Huawei cloud. Enterprise customers will consume cloud services such as video, computing. In future, trend will be enterprises will move to hybrid cloud and public cloud will take a major share. Huawei cloud provides compute/storage/networking to enterprises and government. 200K x86 servers in Huawei cloud. Revenue with external customers - won't share it with you - maybe .
AI chipset question: We don't position chipset as a standalone business - won't sell to external customers. Will be used to differentiate Huawei products. Smartphone - we use multi-vendor strategy always; in other worlds; have multiple Qualcomm, NTK and others. Remain committed to multi-vendor strategy. Don't want vendor lock-in, however. If we only have one vendor, what might happen to our smartphone business, Xu asked.
Enteprise business growth? Declined to comment on specifics, but said he encourages each to grow rapidly.
How do customers get to 86% AI usage (the question was asked by an audience member by incorrectly referring to the statistic that was made during the keynote - specifically, Huawei said AI experimentation will be 86%, not AI usage)? Will give clearer answer at Huawei Connect 2018. For now, can share that we will use AI on ourselves first, then help customers on various functions such as finance, human resources, networks, etc.
Supply chain - will you acqire your suppliers? We do joint innovation with suppliers to meet Huawei's needs; push multi-source strategy, however. Will not invest in DRAM, display, flash.
5G wasn't mentioned much in the presentation, why? We don't have as high expectations as some others; 5G is just one of many products we offer and is just a natural evolution from 4G. You don't have a fundamental difference between 4G and 5G - consumers just see faster speed and lower latency. LTE already support autonomous driving. Past couple years, governments have regarded 5G as too important. June 2018, will only address eMBB - faster speeds. 2019 - will have fully 5G compliant system that does low latency. 4G is pretty robust; we don't see 5G as a national coverage network - it'll just focus on city centers. However, once one carrier announces 5G, then all others must. 50% of Chinese have wireless connection capable of 4K but there are still no 4K stations.
2H18, end to end 5G solutions available. 3Q19, will launch 5G capable phones.
Share trends for Huawei at operators. Revenue growth of telecom services is a challenging topic. This revenue growth topic is why titan operators express concern about moving to 5G; instead, Huawei thinks moving to improved intelligence will assist operators. Video will become more and more important as telcos become media companies too.
Will AI become a privacy concern? Any technology has double-sided effects. With AI, some believe it can be dangerous. Xu believes in the wisdom of man. Look back to history of mankind, and our humanity can do same for mankind.
Will Huawei find alternate suppliers for data center products? (Xu also smiled about this question before answering). Today, Intel is dominant player. Our point of view, we look forward to more diversified landscape; but we work with Intel mainly now.
At NetApp's analyst meeting today, CEO George Kurian sees opportunity in selling HCI (introduced F2Q18, 4-5 months ago), AFA, share-taking in SAN, and public cloud software and services. Every large customer NetApp talks to, according to Kurian, is using multiple cloud service providers and/or SaaS services and most are using the hybrid cloud, which means using workloads both on the customer premises and public cloud. According the company's marketing and sales executives, the company's sales and marketing strategy is focused on leveraging the company's entrance to the cloud services software market.
Substantial future announcements that were made by NetApp:
Summary of presentations
Joel Reich, EVP Products and Operations discussed trends in data center flash:
• NVMe over Fabrics
• Storage-class memory as cache
• Persistent memory in server
• Quad level cell NAND
Reich made some interesting comments:
Brad Anderson, SVP and GM, Cloud Infrastructure BU, said that NetApp’s “Converged” (selling NetApp storage with non-NetApp servers) FlexPod business is now at a $2B run rate and >4,000+PB shipped. The company recently initiated a Fujitsu partnership on March 26, 2018. Anderson also said that NetApp’s Hyperconverged product, which has only been selling for the past 4-5 months, hit its financial targets in the first full quarter of shipments. He also said that the HCI product is based upon on recently-acquired SolidFire technology and conceded that the company is hiring people with virtualization capabilities to further augment the product line. HCI customers that were discussed during this presentation were: ConsultelCloud (Austrialian SaaS company) and Imperva (security company).
Anthony Lye, SVP Cloud Data services BU, joined a year ago and is responsible for the company’s efforts to build software that runs on and with public cloud services. He describes this software as one that operates above the storage layer, to allow customers to manage their data, whether in cloud, SaaS applications or on premises. It offers backup, disaster recovery, and for securing data, and then binds those services and data in context of applications and business policy using the orchestration Engine. OnCommand is product name. The underlying technology NetApp uses is called ONTAP Data Management, which Mr Lye explained was separated from its engineered systems (hardware) and port it to public clouds five years ago. We remember when NetApp announced its plans to separate ONTAP as a software for the cloud at its analyst meeting a few years ago when Kurian took over as CEO.
Lye explained that “later this year,” NetApp will release cloud-based OnCommand performance management/monitoring tool to manage workloads in hybrid cloud environment.
Henri Richard, EVP Worldwide Field and Customer Operations said "Cloud is soon to be GA.” Richard explains this as its “Cloud Volume” product. Richard explained that what is new this year is the hyperscaler relationships, starts a demand creation engine for the sales organization.
Jean English, SVP Chief Marketing Officer said the company will focus on “cloud first” to reach new “global” buyers (e.g. multinational organizations), will lead with HCI and Cloud to enterprises.
Ron Pasek, EVP and CFO explains that FY18 is almost over and the company is beating FY18 plans (low-single digits growth), driven by flash. The CFO said that the new accounting rule, ASC 606 impact to guidance will be immaterial to the P&L , though will result in slightly higher product revenue recognition.
Additionally, Pasek said that a year ago, he said revenue growth will be “low-single digit growth” (FY18-20) and now he is saying “mid-single digit growth,” driven by Flash, HCI, cloud data services. Pasek said that in FY19 cloud data services will represents one point of growth. (As an aside, we calculate FY19 cloud data services revenue, using the “one point of growth” metric at $60M, based on the latest quarter of total revenues, F3Q18 which was $1.52B, multiplied by 4, then multiplied by 1%). So, cloud services revenue is expected to grow to FY19 of $60M and reach FY21 targets of $400-600M. The company declined to state its FY18 cloud data services revenue when asked by the audience, so we take it that it is small.
Mavenir held a 'virtual' analyst meeting - essentially a webinar - that conveyed the company's efforts on several new initiatives: (a) SMB UCaaS - small business unified communications, (b) Messaging as a Platform (MaaP), (c) Monetization of messaging, (d) Security Solutions (relating to toll fraud, for instance), and (d) xRAN/Open RAN/Cloud RAN. Pardeep Kohli, CEO of Mavenir explained that the company is at a $450M revenue rate.
Small Business Unified Communications - its mobile UCC service can be deployed as a service or on premise. The company sells through its service provider partners. Messaging as a Platform - the company sees its capabilities in RCS messaging as a means to enabling branding, chatbots, sale of digital goods, and enriched calling. A month ago, at Mobile World Congress in Barcelona, we saw the Mavenir MaaP system in work - it looks nothing like what is available from traditional telecom equipment vendors. Mavenir, it its re-imagined MaaP offering, looks more like an "OTT" company than a traditional telecom supplier. We hear the refrain from service providers - show us new sources of revenue; well, this system from Mavenir looks to us like a new source of revenue. On a secondary basis, to the extent Mavenir is successful in getting its MaaP system deployed at service providers, we think this can help the company grow its presence with other products.
Mavenir recently acquired Acuto, a monetization specialist company for messaging. Susie Kim Riley, Aquto's founder and CEO, now part of Mavenir, presented the company's offerings. Riley showed off the Sponsored Data system of enabling brands to provide mobile data connectivity to smartphone consumers. She explained how Facebook, Google, Alibaba and Baidu offers free access to consumers in various countries to consumers who don't pay for cellular data. The company discussed that its customers typically pay for data using Zero-Rating (where a marketer can enable users to download/use apps or browse specific sites without using any of their data plan) or Data Rewards (where a marketer can reward users with additional data buckets for taking a specific action like downloading an app). Mavenir showed how Banko Azteca offered both Zero-Rating and Data Rewards to engage with customers and had strong results. We asked the company how long it takes to get a SP up and running and Mavenir said it takes a couple months, depending on the billing system integration.
The company discussed its Radio Access Network and Telecom Core products, too. Mavenir is making "Whitebox LTE" available by integrating its Cloud RAN technology with a Universal Customer Premises Equipment (uCPE) offering, which allows enterprise-deployment of CBRS and LTE Licensed frequencies. We asked the company after the analyst meeting whether its 5G Core (5GC) is available today, and management said, in fact, it is. We think this means a pre-standard version of its 5G Core is ready for customers to take delivery of today, and architecturally, it is similar to its control user plane separated system that performs EPC for customers. Additionally, the company highlighted a new capability that it has made available to the market that is calling its breakout gateway - the goal of this product (technically, a vSAEGW) is to allow mobile operators who carry lots of video traffic for mobile customers to offload this traffic to other operators, thereby reducing carrying costs. I'm not sure how this impacts the quality of video traffic and what is effect on churn might be, but the company claims that by using this breakout gateway, it calculates that it can reduce spending on EPC/5GC by as much as two years.