ACG Research

ACG Research
We focus on the Why before the What
Showing posts with label Tim Doiron. Show all posts
Showing posts with label Tim Doiron. Show all posts

Tuesday, May 3, 2016

Trends and Directions in Data Center Interconnect: A Survey of Optical and Packet Mode Networking Practices

This ACG Research report investigates the trends and directions that service providers are taking in the deployment of their data centers and specifically their interconnection. It provides insight into existing data center practices as well as future practices. The survey report connected directly with service providers utilizing data center interconnect equipment or planning to deploy such equipment to interconnect their data centers in the next 12 months. Respondents included Network Service Providers, Cloud Service Providers, Internet Content Providers and Inter-eXchange Providers in the APAC, EMEA, NA and LAC regions. 

“We confirmed our previously held belief that the number of data centers will grow approximately 60% between now and 2019,” says Tim Doiron, principal analyst, Intelligent Network Services. “We determined that the types of products (SFF or multi-slot chassis) and desired product attributes differ based upon the service provider segment. Data center optical reach demonstrates a bimodal distribution with maxima below 30km and above 600km distances. Traffic drivers for DCI bandwidth also differ by service provider segment.”

Contact kgrenier@acgcc.com to purchase the report and 30 minutes of analyst time.

     Tim Doiron
     www.acgcc.com

Monday, April 18, 2016

PAM-4 or Coherent DWDM for DCI?


At the March 2016 OFC conference, Inphi announced its delivery of a 100G, QSFP28, PAM-4, pluggable transceiver with 80km reach. PAM technology has been utilized for 100G transmissions (Inphi is a specialist in this area) before but at much shorter distances. Pulse-amplitude modulation (PAM) is an analog transmission scheme similar to NRZ but with multi-level signaling, with PAM-4 utilizing four levels to signal one of four possible symbols (2 bits per symbol). During the announcement, Microsoft also publicly announced that it will begin sourcing the pluggable PAM-4 technology from Inphi for interconnection of its regional, metro-distributed data centers, which by definition are within 70km of each other. Coherent technology will continue to be used elsewhere. The metro-distributed data center deployment model builds and interconnects a number of smaller data centers within a metropolitan area instead of deploying a single hyperscale data center in the region. Microsoft also divulged that it was their intention to turn up all 40, 100G wavelengths at one time (4Tb/s with each carrier occupying 100GHz channel spacing) on a fiber pair, utilizing all available colors in the fixed-wavelength portfolio. 

Some at the conference reacted to the Inphi/Microsoft announcement by declaring the obsolescence of existing optical DCI/coherent DWDM solutions. Although the Inphi/Microsoft announcement is exciting news, ACG thinks the PAM-4 technology is far more complementary to existing coherent DWDM solutions than competitive for multiple reasons. 


Figure 1. Optical Reach for 100G Technologies 

Reach. The PAM-4 solution covers a portion of the optical reach needed to interconnect data centers. Below 10km, IEEE 802.3ba 100G pluggable optics are readily available with 100GBASE-LR4 supporting 10km reach in a QSFP28 package for cost-effective point-to-point connectivity. The 100GBASE-ER4 specification for 40km reach has been more challenging for optics suppliers to deliver and remains either in larger packages (example, CFP, CFP2) or in nonstandard formats, meaning non-interoperable across vendors. So where does the PAM-4 technology fit? In general, its initial fit appears to be in the <40km range as an alternative to existing, suboptimal pluggable solutions. We believe there is limited overlap with coherent DWDM solutions in this range. The solution also plays in the 40–80km range as an alternative to optical DCI/coherent DWDM solutions for some deployment scenarios. 

So, based solely upon reach, a logical question is how much of the optical DCI/coherent DWDM market is covered by 40–80km? ACG Research recently completed a worldwide survey of data center service providers, including network service providers, cloud service providers, Internet content providers and Internet eXchange providers. This research will be available in a published report later this month (April). One of the questions we asked the service providers was the proportion of optical reach needed to cover their data center interconnections today and in 2019. What we found is that service providers on average believe that 30–80km optical reach is needed for approximately 30% of their data center interconnections. The results indicate a modest increase between today and 2019. Based upon this preliminary research, we have a sense of the addressable optical DCI market for this technology. However, we also believe that service providers will consider at least three other factors in making their DCI deployment decisions.


Figure 2. Data Center Interconnect Optical Reach 

Operations. Every data center deployment is not like Microsoft’s plan for metro-distributed data centers, which is to turn up all 4Tb/s of connectivity in a point-to-point fashion on day one of data center activation. By deploying all 40 wavelengths at once, Microsoft could reduce the incremental cost per wavelength of deploying dispersion compensation on the fiber, which is required for PAM but not for coherent DWDM solutions. Dispersion compensation costs include both the capital equipment as well as the operational costs associated with installing and tuning the compensators. Microsoft also avoids the operational complexity of deploying fixed wavelength pluggable optics incrementally, where inventory and on-site resources are required every time a change or a wavelength addition is needed. 

Other service providers that have existing metro optical networks may not want to deploy in this manner. They may not want the added complexity of dealing with dispersion compensation for PAM deployments. Some may want to utilize existing metro optical infrastructure and/or deploy in a mesh architecture. Still other service providers may not have the same visibility as Microsoft with regard to their data center connectivity needs. They may need to be more agile and utilize a pay-as-you-go/pay-as-you-grow deployment model where they add interconnection capacity over time and in alignment with their data center compute/storage capacity and revenue generation. An incremental deployment model is just more operationally complex with fixed-wavelength pluggable optics. 

Fiber Scarcity. When fiber is scarce or expensive, fiber optic transmission efficiency (bits per Hz) increases in importance. The PAM-4 solution delivers an efficiency ratio of 1 with 100Gb/s transmission occupying 100GHz channel spacing. 16-QAM coherent DWDM modulation offers 200Gb/s in 50GHz channels or an efficiency ratio of 4. Recent flexible grid implementations have an even greater efficiency ratio approaching 7. If more than 4Tb/s of connectivity is needed and incremental fiber is scarce or expensive, service providers may need to utilize the more efficient coherent DWDM system to squeeze more bandwidth through their limited fiber resources.

Programmability. Fixed-wavelength pluggable optics do not advance the broader drive toward a programmable, agile, SDN enabled optical underlay. SDN and NFV are changing all aspects of the ICT industry, including optical solutions. Service providers are looking to utilize intelligence, automation and programmability to reduce operational costs and ensure that network resources adapt to changing business and networking conditions across protocol layers, including optics and IP. Many demonstrations at OFC utilized SDN control and service automation combined with a programmable optical layer to showcase network efficiency and adaptability. The ONS 2016 conference had similar demonstrations with ONOS and ODL controllers programming in near real-time optical and IP networking infrastructure. 


Figure 3. Example of a Mixed Technology DCI Deployment 

The Inphi PAM-4, QSFP28 solution is an exciting achievement and addresses a very real need in the sub-80km 100G market. We believe the solution is actually far more complementary than competitive to existing optical DCI/coherent DWDM solutions. Most service providers will utilize an all-of-the-above approach to their 100G DCI deployments just as they did before with dark fiber, IEEE pluggables and coherent DWDM options. PAM-4 meets the needs of data center operators, such as Microsoft, that intend to turn up 4Tb/s of transmission capacity in a point-to-point fashion between data centers in a ~70km metro-distributed network. However, if a provider needs longer reach or more than 4Tb/s per fiber pair or an incremental growth operational model or if a service provider is looking to advance its programmable, SDN enabled network, then a tunable, coherent DWDM solution is a better fit. PAM-4 or coherent DWDM for data center interconnections? Yes!


Click for more information about Tim Doiron and his recent articles.

     Tim Doiron
     www.acgcc.com

Friday, April 8, 2016

Infinera Delivers the Multi-Terabit Infinite Capacity Engine

Infinera revolutionized optical integration with the introduction of its industry leading 100G Photonic Integrated Circuit (PIC) in 2005.

In 2011 the company followed with the introduction of a 500G PIC and coherent digital signal processing (DSP) technology.

At the OFC Conference in March 2016, Infinera once again pushed the limits of optical integration with the debut of its multi-terabit Infinite Capacity Engine.

The Infinite Capacity Engine is a family of next-generation optical subsystems consisting of fourth-generation photonic integration with advanced coherent signal processing, software defined networking-enabled sliceable photonics architecture and Layer 1 encryption.


For more information about ACG's market impact service, contact sales@acgcc.com.

     Tim Doiron
     www.acgcc.com

Monday, April 4, 2016

Optical Infrastructure and Optical DCI Finish Strong in 4Q-2015 and Look to Future Growth

Optical DCI contributed over $1B in 2015 with total Optical infrastructure finishing the year at $13.3 billion

ACG Research has released its 2H-2015 worldwide Optical infrastructure and worldwide Optical Data Center Interconnect (DCI) forecast. The forecast period runs through 2020.  The worldwide Optical infrastructure market is predicted to grow from $13.2 billion in 2015 to $17.8 billion by 2020. Purchases of Optical DCI equipment are expected to grow from $1.03 billion in 2015 to $4.3 billion in 2020. ACG Research predicts growth in all geographic regions including EMEA where total optical networking revenues have been flat to down over the past several years.   

Optical infrastructure demonstrated its usual seasonality throughout 2015 with Q2 and Q4 being the strongest calendar quarters.  After a down Q3, Q4-2015 saw growth in both Metro (POTS + Metro DWDM) and Long Haul optical segments at robust 17.5% and 19.5% q-q rates, respectively.  For the year, Metro optical produced 5.2% growth while Long Haul delivered 7.9% for a combined High Speed Optical (HSO) annual growth of 6.4%.  When combined with the 14.3% decline in legacy optical infrastructure spending, total optical infrastructure managed positive 1.2% growth in 2015 to finish at $13.3B.  Looking forward, ACG Research anticipates 6.6% Long Haul optical CAGR and more than 10% Metro optical CAGR over the forecast period. 


Optical DCI equipment revenue exceeded $300m for the first time in 4Q-2015 to contribute more than $1B to the Optical infrastructure market for the year with an annual growth rate in excess of 40%.  Optical DCI revenue is projected to grow at a 33.1% CAGR from 2015 to 2020. The fundamental underpinnings of DCI growth remain strong:  annual data center bandwidth growth, increasing service requirements for data center interconnectivity and increases in the total number of data centers worldwide.  Over the forecast period, ACG predicts the Metro Optical DCI growth rate will exceed Long Haul Optical DCI and SFF Optical DCI appliances will grow at a faster rate than multi-slot Optical DCI chassis-based solutions, although multi-slot chassis solutions will remain slightly dominant throughout the forecast period.  A series of publicly announced new entrants to the SFF Optical DCI appliance market including Ciena Waveserver, Fujitsu 1Fininity, Adva CloundConnect, Cisco NCS 1002, Coriant Groove G30 will join the market leading Infinera CloudXpress in 2016 and keep downward pressure on prices. 

Additional growth drivers beyond DCI for Optical infrastructure over the forecast: accelerating 100G/200G+ coherent optical upgrades, mobile front-haul, 5G mobile backhaul and bandwidth expansions, multi-layer encryption/security and transport/multi-layer SDN.  

     Tim Doiron
     www.acgcc.com

Tuesday, December 8, 2015

SDN/NFV: Intelligent Transport Networking

Tim Doiron, principal analyst, Intelligent Transport Networking, ACG Research, leads an SDN/NFV panel at Layer123 SDN & OpenFlow World Congress in Dusseldorf, Germany.  Tim introduced the panel participants and shared some of the recent findings of ACG Research as part of the panel kickoff.  In working closely with a number of customers, ACG Research has found that through software automation, service providers cannot only accelerate new service introduction, but also substantially increase revenue.  With more rapid service introduction, service providers can expedite time to revenue, enable reduced services pricing, thus attracting more trial customers and finally obtain more paying customers faster.  In total, ACG Research analysis indicates that this virtuous software-enabled cycle can deliver as much as 400% higher revenue generation over a five year period vs. today’s highly manual new-service introduction processes. 


Click for more information about Tim Doiron or to discuss this topic contact Tim at tdoiron@acgcc.com.  

Tuesday, November 10, 2015

Migration of Services to the Data Center Driving Optical DCI Growth

ACG Research has released its Q2/2015 worldwide Optical Data Center Interconnect (DCI) market share analysis as well as its 2014–2019 worldwide forecast for Optical infrastructure platforms purchased by service providers for use in data center interconnect applications. Optical DCI product segmentation includes products designed for both long-haul and metro deployments, as well as a parallel view of the market based on large-scale multi-slot chassis platforms and small-form factor (SFF) optical appliances. The top three optical DCI suppliers worldwide in Q2/2015 are Ciena, Infinera and Alcatel-Lucent, respectively.

Purchases of Optical DCI equipment are expected to grow at a compound annual growth rate (CAGR) of 44.9% during the forecast period from just over $1.1 billion in 2014 to $4.7 billion in 2019. Sales of metro DCI platforms (supporting DCI connections up to 150 km) will continue to dominate over long-haul; both metro and long-haul will experience considerable growth at 51.5% and 24.6% CAGRs, respectively. Throughout the forecast period, the Americas and specifically North America remain the dominant geographical location for Optical DCI. EMEA and APAC regions demonstrate considerable optical DCI growth, but each remains about half the size of the Americas market.

Although the majority of Optical DCI deployments to date have been with multi-slotted chassis products, small-form factor Optical DCI appliances are entering the market at a rapid pace, led by Infinera’s two rack-unit (2RU) Cloud Xpress, which debuted in late 2014. Recent announcements from other vendors in the optical appliance category include Ciena’s Waveserver™ and Fujitsu’s 1Finity™ platforms. Adva also recently debuted its FSP3000 CloudConnect™ platform, though Adva is espousing a modular, 4RU chassis as “right-sized” for Optical DCI applications. Expect to see more product announcements in the future for this fast-growing product segment as revenue is projected to approach parity with large multi-slot chassis solutions in the last year of the forecast period.

“Uptake of Optical DCI is being driven by the migration of services to data centers and the cloud as service providers simplify deployment models and accelerate delivery of new and differentiated services,” says Tim Doiron, practice lead for Intelligent Transport Networking at ACG. “New and expanded data center deployments are being driven by a variety of service providers including Internet content providers (ICPs), network service providers (NSPs) and interexchange providers (IXPs) as well as enterprises themselves. As more functions become automated and virtualized, the need to interconnect data centers for capacity, resiliency and versatility will continue to grow and increase the need for reliable, cost-effective, high-speed data center interconnections.”

For more information about ACG’s data center interconnect services contact tdoiron@acgcc.com or info@acgcc.com.

Click for more information about Tim Doiron or to discuss this topic contact Tim at tdoiron@acgcc.com.

Thursday, October 22, 2015

SDN & Multi-layer Transport SDN: Notes from Layer123 SDN OpenFlow World Congress

This year’s Layer123 SDN OpenFlow World Congress in Dusseldorf, Germany, was quite an expanded event from last year with over more than 1,500 people registering.

There was a great mix of presentations from equipment suppliers, services providers and open source organizations at the event. SDN and NFV were, of course, top of mind at the event. The number of SDN and NFV PoCs and trials continue to grow rapidly, but live commercial deployments outside the data center remain elusive. Our ideas and thinking about the application of this technology in our networks has, however, matured. The focus has shifted, correctly I believe, from minimizing capital costs with COTS hardware to agile revenue generation via network automation and programmability.

Although many challenges remain, the single biggest barrier to mass SDN commercial deployment is operationalization of the technology. It is not just commissioning either. A virtualized and programmable network must still be operated and managed throughout its life-cycle to meet changing networking demands and customer service level agreements. In one conversation with an equipment manufacture, we discussed the simple scenario of a fan failure in a server running multiple VMs and VNFs. Who would know of the failure? How would they know and when would they know? Part of the beauty of an NFV environment is that the VM/VNF can simply be moved to other physical machines. However, financial considerations will always dictate that there is a limit to the number of physical machines (COTS or otherwise) installed in a service provider network. The underlying physical network will have to be maintained and failures addressed lest they eventually lead to poor network performance and customer satisfaction.

The fact that there was broad acknowledgment about the need to close the operational gaps is encouraging and a major step toward increasing commercial deployments.

Multi-layer Transport SDN was another topic that generated a lot of chatter in both Layer123 sessions and at a lunch-time debating table. Is multi-layer only through Layer 2 or 2.5? Or does it involve Layer 3 and IP?

After some discussion, the general consensus emerged that in order to maximize the value of an agile SDN-enabled network, multi-layer SDN and associated path computation must be Layer 0-3. The value of a multi-layer control plane is significantly diminished if IP is not a part of the solution. Independent fault detection and recovery mechanisms (think path computation) is exactly what we have in today’s networks with the packet-optical layers doing their own detection and restoration while IP executes its own Layer 3 detection and restoration mechanisms with protocols such as BFD and EMCP. Break a fiber in a network and all layers work almost completely independently to restore paths and services at their respective protocol layer.

With SDN and centralized control, we have the opportunity to ensure that wavelengths, ports and paths are coordinated and utilized for maximum efficiency. We can simplify our networks and drive out complexity and operational costs. Must a supplier’s controller and path computation element (PCE) contain Layer 0-3 functionality? Not necessarily. The hierarchical nature of SDN control means that hierarchical-PCE across multiple PCEs is a viable option. Packet optical suppliers could focus on Layer 0-2 PCE but then interface in a hierarchical manner with a Layer 3 PCE partner/supplier. Alternatively, a monolithic Layer 0-3 PCE is also possible but might require tighter coordination and integration than an equipment supplier may want to pursue. Either way, packet optical suppliers need to drive their PCE thinking from a Layer 0-3 perspective if we are to simplify the network, improve equipment utilization/efficiency and create agility for the future.

Click for more information about Tim Doiron or to discuss this topic contact Tim at tdoiron@acgcc.com.


   Tim Doiron
   tdoiron@acgcc.com
   www.acgcc.com