A critical component for new telco services - · PDF fileFuture-proofing ADCs A critical component for new telco services Sponsored by

Future-proofing ADCs A critical component for new telco services

Sponsored by

2© 2015 Ovum. All rights reserved. www.ovum.com

Executive summary .........................................................................................................................................4

Market overview ..............................................................................................................................................5

New areas of growth .....................................................................................................................................5

Enterprise services .......................................................................................................................................5

The importance of the CSP data center .......................................................................................................5

Enabling services at the network edge ...........................................................................................................6

Multi-tenancy ................................................................................................................................................6

CGNAT ...........................................................................................................................................................7

Firewall/VPN .................................................................................................................................................7

Load balancing ..............................................................................................................................................7

Scalability and cost efficiency of ADCs .........................................................................................................7

TCO analysis ....................................................................................................................................................7

Methodology ..................................................................................................................................................7

CSP scenarios ...............................................................................................................................................8

Conclusions....................................................................................................................................................10

Citrix Brings New Levels of Efficiency with New ADCs .................................................................................11

Contents

© Copyright Ovum 2015. All rights reserved.

The contents of this product are protected by international copyright laws, database rights and other intellectual property rights. The owner of these rights is Informa Telecoms and Media Limited, our affiliates or other third party licensors. All product and company names and logos contained within or appearing on this product are the trademarks, service marks or trading names of their respective owners, including Informa Telecoms and Media Limited. This product may not be copied, reproduced, distributed or transmitted in any form or by any means without the prior permission of Informa Telecoms and Media Limited.Whilst reasonable efforts have been made to ensure that the information and content of this product was correct as at the date of first publication, neither Informa Telecoms and Media Limited nor any person engaged or employed by Informa Telecoms and Media Limited accepts any liability for any errors, omissions or other inaccuracies. Readers should independently verify any facts and figures as no liability can be accepted in this regard - readers assume full responsibility and risk accordingly for their use of such information and content.Any views and/or opinions expressed in this product by individual authors or contributors are their personal views and/or opinions and do not necessarily reflect the views and/or opinions of Informa Telecoms and Media Limited.


About the author

Dimitris Mavrakis

Dimitris Mavrakis is a Principal Analyst with Ovum. He is part of the Intelligent Networks team where he covers a range of topics including LTE, LTE-A, 5G, SDN, NFV, WiFi, IoT, network APIs and identifying how under-the-radar technologies may disrupt or improve the mobile value chain.

Dimitris is also actively involved in Ovum's consulting business and has led several projects on behalf of Tier-1 operators and key vendors.

Dimitris has over 10 years experience in the telecommunications market. He has a strong background in mobile and fixed networks and an in-depth understanding of market dynamics in the telecoms business. In the past, Dimitris has worked as a project leader to perform network measurements and road tests. In his academic career, he has led a team of researchers to produce pioneering research and acclaimed publications.

Dimitris holds a PhD in Mobile Communications and an MSc in Satellite Communications from the University of Surrey.


Executive summaryThe telecoms market is becoming an increasingly hostile and competitive environment, where regulation, new OTT entrants, saturation, and the economic environment are eroding the telco operators’ profit margins. On the other hand, there is heated discussion in the industry about the need for new technologies and platforms that will help these same operators operate a much leaner network and business, which will also lead to new business opportunities. Network virtualization, the evolution from proprietary to commoditized hardware, software, service agility are some of the trends that are reshaping and, in some cases, disrupting the established operator business.

Over the last two years, there has been a shift in telecoms operator growth strategy away from the consumer market and towards the B2B sector. While operators are being disrupted in the consumer sector by Internet companies, they have the opportunity to be disruptors themselves in IT services and, specifically, cloud computing.

Operators are at very different stages in their efforts to develop the enterprise market. Some, particularly national incumbents, have mature enterprise businesses and face the same competitive challenges as in the consumer sector. Others, particularly mobile-only operators that have only started building out portfolios of enterprise services in the last five years, are seeing revenue growth potential but are starting from a low base. Enterprise mobility is one of the hottest B2B sectors and plays to their strengths.

Success in the B2B sector requires a network that can cope with enterprise requirements and the Gi-LAN has been identified as a ripe environment for service innovation in today’s operators. The Application Delivery Controller (ADC) is a critical piece of the service delivery puzzle, albeit one that is not glorified or widely discussed. ADCs allow the uninterrupted operation of many operator data services and have evolved from having singular capabilities to offering enterprise services (VPN, firewall), load balancing between service chains, and CGNAT for IoT. The ADC is considered a key component for enabling enterprise services and business in new areas.

As illustrated in our TCO study later in this white paper, selecting the right ADC at the right cost will create the environment operators need to maintain growth in these new areas of business. Although operators may choose an ADC platform due to legacy relationships, the increase in required capacity and a difference in cost may make the decision turn towards the more cost-effective platform. Our TCO study compares two ADC platforms from Citrix and F5 and concludes that the TCO savings are significant in high-bandwidth deployments. Given that ADC platforms are being continuously upgraded, Ovum expects operators to be offered new cost-effective platforms that will be able to cater for future applications.


Market overviewThe telecoms market has become a hostile and increasingly competitive environment for all communication service providers (CSPs). Even Tier-1 CSPs that are now planning to evolve to a digital service provider are finding out that revenue growth is a considerable challenge when competition within the existing value chain and new entrants – including Web giants – for basic and advanced communication services is heating up. LTE, LTE-Advanced, FTTx, and VDSL are providing some organic growth for the traditional access-based business models but CSPs need to plan ahead and implement new strategies to avoid becoming simple dumb pipe providers.

Service revenues of top-20 CSPs are declining, and the operational expenditure (opex) accounts for nearly 60% of these earnings. As a result, most CSPs are finding it difficult to reinvest into their business to drive top-line growth or to invest in long-term transformation initiatives that will provide benefits for many years. A common strategy across all service providers now is to redefine operational and business processes in order to enhance the end-user experience, increase customer retention, launch new services in established and new segments, and become a more valuable partner.

New areas of growthThe Communications Technology (CT) environment is merging with Information Technology (IT) to create what is now being referred to as ICT. As the boundaries between the two distinct value chains begin to fade, vendors in both these markets are finding new opportunities outside their traditional reach. Moreover, buyers of technology from these two markets are finding that their needs can be fulfilled by new entrants, including much smaller companies. For example, a contact-center business can now buy voice capabilities from a Web company, while a large enterprise can buy cloud services directly from a CSP. This new ICT value chain is an environment where services are created – and decommissioned – in a much quicker and more agile way.

It is no secret that CSPs are facing considerable difficulties to compete in this rapidly changing world. Despite their shortcomings, CSPs now

have the opportunity to enter new and interesting areas and take advantage of their existing captive user base to launch new services: These new areas include the Internet of Things (IoT) and M2M communications, enterprise verticals, services for small to medium-sized enterprises (SMEs), and cloud services for businesses.

Enterprise servicesOvum’s current research indicates that leading telcos are targeting hybrid and private cloud services, looking to migrate clients from hosting and co-location. They are busy striking partnerships with the public cloud giants for secure cloud-interconnect network services, and are positioning themselves as partners for enterprises that wish to move workloads across public, hybrid, and private clouds on demand. Although systems integrators are doing this too, we believe this is a realistic strategy for telcos – at least for telcos with ICT experience and assets or those willing to buy them.

The importance of the CSP data centerMost of the enterprise services discussed previously are largely enabled by data centers and most CSPs are now spending significant amounts of capex to expand their data-center (DC) footprint to offer new services. CSPs have invested considerable capex to deploy mobile and fixed broadband networks over the past few years, making the years to come likely to be relatively slow in terms of capital intensity (capex/revenues). However, Ovum estimates that the overall spend between 2014 and 2019 for CSPs will be $2tn – and a large percentage of this expenditure will be driven by data-center expansion.

Regardless of the service and revenue opportunity, Tier-1 CSPs are now finding out that they have two major priorities in terms of technology and services: create an environment where it’s easy to innovate; and create new cloud-based services while making service delivery more efficient and agile. For mobile networks, the Gi/SGi LAN is an environment where service innovation can happen and also the space where the data center, the Internet, and the core network interface. As such, we expect that Gi/SGi LAN to be a key component of CSP services in the coming years and telcos will need to pay more attention to the technology they deploy in this particular network domain.


Enabling services at the network edgeThe network edge represents a considerable opportunity for creating new services and fostering innovation. In the case of mobile networks, there are two areas where the CSP network interfaces with the outside world: at the base station, and at the Gi/SGi interfaces. There are industry initiatives being developed for adding intelligence at the base station and other network edges that are close to end users (including the ETSI Mobile Edge Computing initiative) but these are still under development and are not expected to be deployed for at least a few years. The potential for innovation and new services (e.g. augmented/virtual reality, location-based services) is great at the mobile edge, but CSPs do not have a clear business model for this family of technologies yet, which would be a major requirement, especially as the cost to deploy new functionality at the base station is very costly.

On the other hand, the Gi/SGi (please note that Gi is used for both Gi for 3G and SGi for LTE for the rest of the white paper) interface is the gateway to the rest of the Internet and considered a domain where service innovation can be fostered. Moreover, this network edge is not specified extensively in standards – unlike the rest of CSP networks. It is for this reason that the Gi-LAN is considered to be one of the first environments that is likely to be home to new NFV deployments for new services while many NFV Proof of Concepts are assessing different applications in the Gi-LAN, one of them being service chaining. Some examples of service chaining include:

a new malware attack is reported

Delivery Network (CDN) to ensure user experience is maintained

may host applications on the carrier cloud, for example office applications, firewalls and VPNs

An important component in this process is the ADC, which resides between the Internet and the CSP network components outlined above. ADCs are in charge for many functions, the most important of which is to route traffic according to the application. Figure 1 illustrates a simplified Gi-LAN: The ADC is responsible for routing traffic through different service chains in this particular illustration but it is also an enabler for a variety of use cases, outlined below.

Multi-tenancyA CSP’s business may benefit from multi-tenancy through several different ways:

distinct virtual networks with specific SLAs and network routes

CSP cloud and governed by granular SLAs.

In the past, multi-tenancy was supported by different sets of equipment and hardware-driven service chains. Current virtualization efforts aim to decouple hardware from software and thus allow more flexible introduction, scale-in, scale-out, and destruction of functionality for multi-tenancy applications. Multi-tenancy is considered a primary business target for NFV in the Gi-LAN and the ADC is a critical component to distribute traffic throughout these service chains.

Radio access Core network

Enterprise traffic

Internet

M2M/IoT traffic

Physical or virtual functionsGi/SGi LAN

DPI Firewall/

CGNAT

ADC

Media

optimiser

M2M

gateway

Figure 1: Simplified Gi-LAN environment

Source: Ovum


CGNATCarrier-Grade Network Address Translation (CGNAT) refers to IPv4 translation in order to share a limited address pool with several end points. In the context of mobile networks, CGNAT translates local mobile IP addresses to public addresses so that operators can oversubscribe the public address space allocated to them. CGNAT is vital for mobile networks – particularly as smartphone penetration increases and data services become mainstream – and will even become more important as telco IoT and M2M businesses become revenue drivers. CGNAT is also necessary to protect investments in IPv4 equipment even when IPv6 is widely deployed.

CGNAT is a primary use case for an ADC, which performs several additional functions for connecting end points. For example, ADCs allow what is referred to as “hair-pinning”, where end-points are allowed to communicate within the operator network without their addresses being translated to the public domain.

Firewall/VPNADCs are ideally positioned to act either as a firewall or as an end point for remote connections to the telecoms or enterprise network. Their position at the edge of the network and their capabilities make them ideal for both use cases. Several CSPs are using ADCs for both of these use cases, especially as SSL VPNs are widely deployed in the market today.

Load balancingArguably the most important function of an ADC is load balancing, i.e. distributing traffic across the network to several appliances. For example, an ADC is responsible for distributing video traffic across optimization appliances, which may either be physical of virtual. With the advent of virtualization, the use of the ADC may become even more necessary to distribute traffic across service chains or virtualized components according to network policies.

Scalability and cost efficiency of ADCsADCs used in a telecoms environment have stringent requirements for robustness, stability, cost efficiency, and upgradability. There are several ADCs in the market today, although very few of them can scale to the level of Tier-1 CSP requirements.

Mobile operators can choose to deploy many smaller ADCs and scale according to traffic requirements but they soon find out that these smaller ADCs are far too expensive compared with larger platforms. Several vendors offer carrier-grade ADCs that can provide services to millions of users. The following section provides a brief analysis on the economics of ADCs and a like-for-like comparison of vendors.

TCO analysisAs discussed in previous sections, an ADC is a critical operation for the uninterrupted and profitable operation of a telecoms network; especially as new areas are being targeted, including enterprise verticals and IoT. As such, operators have been dimensioning and planning their ADC platforms according to established overprovisioning techniques which use linear projections for ADC throughput and other dimensioning parameters. However, as LTE and LTE-Advanced usage grows, these platforms will be increasingly put under strain, especially as TCP/video optimization, data-center multi-tenancy, traffic steering, and load balancing for NFV become critical.

This section presents a TCO analysis of major ADC platforms available in the market today and could act as a guide to compare vendors on a like-for-like basis. All data for the analysis below has been independently sourced and validated with vendor data sheets or publicly available material – including pricing.

MethodologyThree ADC platforms have been chosen to be benchmarked in this analysis, according to data sheet, pricing, and deployment example availability across the globe. The three ADCs being benchmarked are:

The choice of these specific ADCs is largely driven by the availability of pricing information from public sources, including the Pennsylvania Education Purchasing Program for Microcomputers (PEPPM), a national US program for education IT purchasing, and the Office of General Services (OGS) for the New York State. It is noted that the pricing represented in


this list may not necessarily reflect “street pricing” or what CSP contracts would include, but these are expected to be reasonable relative benchmarks for comparing vendor equipment and comparing TCO calculations.

The three ADCs chosen represent carrier-grade platforms listed in the PEPPM/OGS databases by each vendor. The database includes different configurations for each of these three products and the most basic implementation for each has been chosen for this analysis. Although there are several configurations for these ADCs – for example, ADC for security – these are different for each vendor and each ADC, making a like-for-like comparison impossible. The configuration of ADCs used for the TCO comparison includes CGNAT and load-balancing capabilities in a telecoms environment.

The simple versions of all the ADCs compared here is more likely to provide a meaningful analysis and benchmark in terms of simple metrics, including throughput, HTTP requests per second, and SSL transactions per second (TPS).

A few selected specifications for the three ADCs chosen are listed in Table 1.

An initial comparison illustrates that, although a larger appliance is naturally more expensive, processing capabilities, energy efficiency, and several other parameters are cost-effective in higher use. Although ADCs for smaller-scale and lighter applications are likely to be adequate for low processing and transactional capabilities, CSP use is likely to require several powerful ADCs which may even need to be scaled up from the use cases outlined.

An initial comparison illustrates that in terms of throughput, the largest appliance is optimized in terms of efficiency (see Table 2).

As expected, the analysis above illustrates that a larger platform does provide cost efficiency in demanding applications. It is also expected that virtualization will not provide an adequate throughput and cost efficiency in the short term, due to the demanding processing capabilities of these ADC platforms. The ADCs listed above are equipment that has been optimized extensively in terms of hardware, software, and cost and so are not likely to be exchanged for virtualized components within the next two to three years.

In terms of energy efficiency and cooling requirements, Table 3 illustrates a direct comparison.

As expected, the more powerful ADCs are more efficient in terms of power consumption and heat output. It should be noted that these are maximum power supply ratings and nominal use is likely to result in much lower energy usage. However, CSP are likely to require heavy use of these ADCs, which will lead to energy performance that approximates the maximum power consumption and energy output.

CSP scenariosA current mid-sized operator – for example, a Western European operator with a moderate subscriber base – is likely to require approximately

Table 2: ADC cost per unit throughput and HTTP Layer 7 requests

Cost per throughput unit ($/Gbps)

Cost per HTTP L7 unit ($/RPS)

Citrix NetScaler 25160T $1,125 $0.039

F5 Viprion 2200 chassis + 2 x 2250 blades

$2,090 $0.083

Kemp LM-5400 $1,763 $0.16

Source: PEPPM/OGS databases, vendor datasheets, secondary research, and Ovum data

Table 3: Energy and cooling efficiency of ADCs

Maximum power consumption (W)

Maximum heat output (BTU/hour)

Citrix NetScaler 25160T 717 2,219


640 2,255

Kemp LM-5400 185 629

Note: Standard configuration includes 2 power supplies (2x850W)Source: PEPPM database, vendor datasheets, secondary research and Ovum data

Table 1: ADC specifications and list pricing

Parameter L7 throughput (Gbps)

HTTP Layer 7 RPS

List price ($)

Annual maintenance

fees ($)

Citrix NetScaler 25160T

160 4,600,000 $180,000 $30,600


160 4,000,000 $334,475 $56,860

Kemp LM-5400

10.2 110,000 $17,990 $3,508

Source: PEPPM/OGS databases, vendor datasheets, secondary research, and Ovum data


160Gbps of ADC capacity in early 2015. However, this capacity is forecast to grow significantly in the next few years: If we consider ADC capabilities for a similar operator in 2013, operators then had deployed approximately 70–80Gbps of capacity. Therefore, CSPs of this size are likely to require approximately 360Gbps of ADC capacity for the next few years, but only if our estimates just take a linear growth of the existing services in consideration. If new services (e.g. enterprise verticals, IoT), which require higher capacity for CGNAT, traffic steering, optimization, and load balancing, are taken into account then these estimates are likely to be considered fairly conservative. Moreover, overprovisioning of 100% has been assumed throughout this study: For example, a single ADC providing 160Gbps of throughput is able to service 80Gbps of traffic.

The operator considered here is operating in a well-developed and competitive market with a subscriber base that is near the 30 million mark.

Of these subscribers, half are still connected via a 3G network but a growing subscriber base is LTE-based. The share of LTE subscriptions is expected to radically increase within the next few years, fuelling further growth in the Gi-LAN.

The TCO for CSP scenarios outlined in this section only takes Citrix and F5 ADCs into consideration. It is highly unlikely that Tier-1 CSPs will use the smaller ADCs that are currently in the market, particularly due to the large number of appliances they would need.

Figure 2 illustrates ADC capex when considering the ADC throughput requirements outlined above.

Similarly, power consumption and cooling are included in the opex needed to run the ADCs. Annual maintenance fees, approximately 17% of the upfront cost for these ADCs, are also included in the TCO model.

Cum

ula

tive a

nnual A

DC

capex

($000s)

F5 Viprion 2200 chassis Netscaler 25160T

0

500

1,000

1,500

2,000

2,500

20172016201520142013

Figure 2: ADC capex for a mid-sized Western European CSP, 2013–17

Note: Capex not depreciated but only calculated for any given yearSource: Ovum calculations, PEPPM/OGS databases, vendor data sheets

Cum

ula

tive T

CO

($000s)

F5 Viprion 2200 chassis Netscaler 25160T

0

500

1,000

1,500

2,000

2,500

20172016201520142013

Figure 3: Cumulative TCO (cumulative capex + annual opex) for ADCs, 2013–17

Note: Capex not depreciated but only calculated for any given yearSource: Ovum calculations, PEPPM/OGS databases, vendor data sheets


Figure 3 illustrates that the more powerful platforms provide considerable savings in terms of opex, which increases proportionally to usage due to energy and cooling costs. The higher capex cost also increases maintenance fees, which increase with the number of ADCs deployed.

As illustrated in the Ovum’s calculations, the higher upfront cost of an ADC, coupled with

maintenance and power/cooling costs, results in a much higher TCO, especially as more capacity is required. In the scenario listed above, usage during 2013–15 only requires a single ADC from either vendor. However, both 2016 and 2017 experience significant traffic growth, which cannot be serviced by existing ADCs, thus more equipment is required which increases the TCO significantly.

ConclusionsAs new opportunities for CSPs emerge, there is new pressure to design, dimension, and deploy a network for the long term. Enterprise services, IoT, and new data services are pushing CSPs to migrate from the legacy physical domain to software and virtualization, with a renewed drive to become agile, flexible, and responsive to user demands. The Gi-LAN is a ripe environment for service innovation and most operators (and vendors) are assessing new service opportunities. The ADC is a thus a critical component for evolving to a service-oriented network. Modern ADCs have evolved from just load balancing and now offer a variety of services, which are applicable to enterprise services, LTE networks, telco cloud and IoT.

Although current ADC deployments have focused on linear traffic dimensioning, new services in the short term are likely to require more capacity in the Gi-LAN, followed by a significant growth in the following years when the CSP business model includes cloud offerings and new verticals. Therefore, there is now renewed pressure to deploy ADC platforms that can scale to a high capacity at a manageable cost, rather than deploy smaller platforms that are likely to need upgrades or expansions in the next two to three years. Given that the ADC market is experiencing healthy growth, Ovum recommends that CSPs should assess the longer-term TCO impact of choosing the right ADC as illustrated in the economic comparison outlined above.


Citrix Brings New Levels of Efficiency with New ADCs

Citrix NetScaler is deployed by Tier 1 communications service providers globally in the network core and data center, supporting over 150 million subscribers. It supports some or all S/Gi-LAN subscriber traffic to accommodate growth in video, app, and Web traffic. As the no.1 application delivery controller for cloud and Web service providers, NetScaler is ideally suited for CSP initiatives to deploy automated networks and provide cloud services.

The Citrix NetScaler 25000 series is one of NetScaler’s newest appliances. Designed for the mobile network, with the NetScaler 25000, CSPs can simplify their S/Gi-LAN, scale out network capacity, lower their cost structure, and prepare for NFV.

Leveraging the latest merchant silicon, the 25000’s software architecture supports Layer 7 capacity of 100 gigabits per second with paygrow to 160 gigabits per second and scale-out clustering to more 3.5 terabits per second.

By offering a price-to-performance ratio that is 32% to 65% better than the competition, a CSP can add 50% to 300% more capacity by using the 25000 at the same investment as existing ADCs. They can deploy the 25000 at a superior total cost of ownership (TCO) to refresh existing ADC deployments and/or expand ADC capacity for new or existing applications.

Unlike ADCs deployed in past years, the NetScaler 25000 can combine multiple services such as load balancing, CGNAT, and subscriber-aware traffic steering at high performance. This eliminates the need for a separate set of ADCs and their corresponding maintenance contracts and operational expenses.

Because NetScaler is software-based, it provides feature and functional equivalence across all of its virtual, appliance, and multi-tenant platforms. That means CSPs with the NetScaler 25000 can transition the platform to a virtual, multi-tenant NetScaler 25000, the SDX 25000, while leveraging the same knowledge, tools, and scripts.

ABOUT OVUMOvum is a leading global technology research and advisory firm. Through its 180 analysts worldwide it offers expert analysis and strategic insight across the IT, telecoms, and media industries. Founded in 1985, Ovum has one of the most experienced analyst teams in the industry and is a respected source of guidance for technology business leaders, CIOs, vendors, service providers, and regulators looking for comprehensive, accurate and insightful market data, research and consulting. With 23 offices across six continents, Ovum offers a truly global perspective on technology and media markets and provides thousands of clients with insight including workflow tools, forecasts, surveys, market assessments, technology audits and opinion. In 2012, Ovum was jointly named Global Analyst Firm of the Year by the IIAR.

For more details on Ovum and how we can help your company identify future trends and opportunities, please contact us at [email protected] or visit www.ovum.com. To hear more from our analyst team join our Analyst Community group on LinkedIn www.ovum.com/linkedin and follow us on Twitter www.twitter.com/OvumTelecoms.

ABOUT CITRIXCitrix (NASDAQ: CTXS) is leading the transition to software-defining the workplace, uniting virtualization, mobility management, networking, and SaaS solutions to enable new ways for businesses and people to work better. Citrix networking solutions for mobile operators elevate subscriber quality of experience, economically scale data and control planes, and enable the transition to next-generation networks. Citrix solutions are in use at more than 330,000 organizations and by more than 100 million users globally. Learn more at www.citrix.com.

A critical component for new telco services - · PDF fileFuture-proofing ADCs A critical component for new telco services Sponsored by

Documents