Bringing you the very latest on SDN and NFV developments across the world

Posts tagged ‘networks’

Interview with Shazia Hasnie, Senior Director, Network Architecture and Strategy, Megapath, USA

Shazia Hasnie-2012Informa’s Carrier Network Virtualization conference and exhibition in Palo Alto, USA on 9-11 December 2013 is now only two weeks away. In advance of the show,  we caught up with Shazia Hasnie, Senior Director, Network Architecture and Strategy, Megapath, USA, to find out her thoughts on the growth of SDN and NFV and her expectations for the industry’s future…

Shazia, welcome. Why do you think the idea of SDN and NFV is suddenly taking off?

SH: “Cloud computing has put tremendous pressure on network connectivity to be dynamically scalable and virtualized. In cloud data center environment, the concept of resource pooling with the ability to dynamically control physical and virtual resources is key. Compute and storage to a large extent have been evolved to be virtualized and pooled resources. However, the network remains static and hence presents a bottleneck. In cloud data center environment, networks need to be agile and virtualized and NFV/SDN provides a way to achieve that.

“SDN has promise in the carrier network environment as well. According to the resent Cisco Visual Networking Index survey, the annual global IP traffic will surpass the zettabyte threshold by the end of 2017. This is a lot of traffic to handle which translates to an ever increasing requirement to augment network capacity. To complicate the matters further, carriers have no automated mechanisms to orchestrate the stranded capacity on its circuits to other customers, who may need it. Hence, this expensive resource remains unutilized.

“Thus, CapEx and OpEx to operate telecom networks are enormous while revenue remains largely decoupled from the cost of architecting, scaling and operating today’s networks. Furthermore, the revenue generating services take several weeks to be created, provisioned and activated on these networks.

“The reason the idea of SDN and NFV is taking off is because it is a much needed concept for both telecom carriers and cloud operators. SDN and NFV present the concept of architecting these networks on low cost COTs hardware while the control of network functions remains in a logically centralized and directly programmable software domain.

2) Whereabouts in Telco networks do you see that SDN and NFV will be of the greatest benefit? Where will we see the first deployments?

SH: “Complete abstraction of control layer from the underlying hardware would take a few years at its best to take place. In the meanwhile, a subset of benefits of SDN would be realized in a software defined service layer. Thus, OSS/ BSS systems in a telco network would most likely be the first to get logically centralized and programmable via a service automation and orchestration layer. This software defined service orchestration layer may interface with the existing OSS/BSS systems and provide an integrated and logically centralized view and control of the network. The other area where SDN and NFV concept would take off in telco networks would be the network edge.

“Most likely, tier 1 operators, globally, will be the early adopters of the SDN technology.”

3) Are there still any major hurdles you see which SDN and NFV have to conquer before adoption can become widespread? If so, what are they and how should they be resolved?

SH: “SDN and NFV technologies are very nascent. There are many pieces of the puzzle which are not in place yet.

“SDN / NFV environment would be composed of physical and virtual resources. Network virtualization poses a great challenge to network state management. Live workload migration and dynamic resource allocation are core concepts of the virtualization technology. However, during the migration, network connectivity, configuration, ACLs, and QoS/CoS should remain consistent and intact. In order to manage server virtualization along with network virtualization, sophisticated management and troubleshooting solutions would be required.

“The existing (non-SDN) networks are here to stay for a long time to come. As an intermediate step, we require multi-vendor service automation and orchestration platforms to provide SDN like capabilities to the existing networks without actual abstraction of control plane from the forwarding plane.

“Also, OpenFlow needs to mature in terms of its switch configuration protocol (OF-Config), extension to optical transport and definition of an open, standardized northbound API.

“Boundary or interface functions are needed to be defined to interface SDN controllers not only to other SDN controllers and domains but also to legacy IP networks.

“SDN would require application-aware routing. The routing / control software that SDN applications would require for advanced functionalities is not ready yet. The IETF I2RS working group was formed in November 2012 to address this problem for the existing distributed routing architecture. The I2RS would allow applications to dynamically modify routing decisions on the basis of application requirements keeping in view the network events, topology and traffic conditions.

“The carrier grade performance in a virtualized networking environment raises some questions and concerns. When complex network functions are running at the software layer, the CPU performance becomes critical. Performance of the current general purpose multicore processors may not fit the bill. New and innovative solutions would be needed to accelerate the performance of these processors.”

4) What kind of timescale are we looking at until SDN and NFV become more commonplace in carrier networks across the world?

“SDN and NFV present a paradigm shift in how networks are constructed and operated today. However, many tier 1 operators are trialing SDN in targeted domains for specific applications. It is given that the existing infrastructure would remain in place for a long time to come while operators strategize to adopt SDN and NFV concepts fully.

“If current state of affairs in the industry is any indication, it can be reasoned that the wider adoption of the technology by telecom carriers would lag behind the wider adoption by data centers at least by a couple of years. However, telecom carriers in general are very aware that their future network architecture would be substantially different from their current architecture.

“It took approximately six years for server virtualization market to grow from 0 to approx. $1B. SDN / NFV adoption would be considerably accelerated then that. A recent report published by Transparency Market Research claims that the global SDN market is expected to reach $3.5B by 2018.”

5) What are your own company’s plans for deploying SDN and NFV?

SH: “Being first to market may not always prove to be prudent. We have to time it right and understand when the adoption of this new technology would add most value to our business and add competitive advantage and differentiation. We are currently in an exploratory phase and doing technical due diligence and impact analysis while formulating our business case. We are interviewing vendors. In our vendor selection, we would prefer a vendor who is committed to open standards, who is very involved in the standardization process and investing in R&D.”

6) What do you expect will be the defining trends within the industry in 2014?

SH: “I believe that enterprises, cloud service providers and data center operators will predominantly drive the SDN market in 2014.

“Cloud computing paradigm is not complete without network being virtualized along with server and storage virtualization. Thus, cloud computing would continue to drive virtualized overlay network technology in the data center environment for its own survival. These overlay network virtualization technologies would continue to be developed and adopted by data center and cloud computing sector.

“However, in the telecom operators’ world, the two key areas would be the virtualization of the network edge and software defined service orchestration also known as OSS/BSS virtualization. The top tier telecom operators, globally, would continue to trial and start to deploy SDN technology for specific use cases, mainly focused around the two key areas mentioned above.

“The mobile network operators would focus on their core and metro networks first, before moving on to RAN for virtualization. An industry survey published by Informa telecoms and media earlier this year showed that approx. 93% of the respondents from mobile operators’ space expect SDN to be implemented in their network in the next 1 to 5 years.

“SDN is colossally disruptive and if the past is any indication of the future, it would enable new vendors and competitors to enter into the arena. Additionally, many channel partner programs will most likely emerge in 2014.

Shazia Hasnie is Senior Director for Network Architecture and Strategy at MegaPath, based in San Jose, California. Being a subject matter expert, she has spoken on the topics of SDN and NFV at various industry events. Shazia has more than 14 years of post-doctoral experience and broad-based expertise in engineering research and innovation, technology and business strategy and management. She holds a Ph.D. in Telecommunications Engineering from the Australian National University.

She will be speaking on SDN and NFV at Carrier Network Virtualization in Palo Alto on 9-11 December 2013.

For further information, please visit www.carriernetworkvirtualization.com

 

OpenFlow wins further support from SDN advocates

SDN-300x225Network management and switching specification OpenFlow is being buoyed by industry support, according to the Open Networking Foundation (ONF), a non-profit organisation dedicated to promoting Software Defined Networking (SDN).

The group recently stepped up development and implementation of the OpenFlow 1.3 protocol, with a conformance testing event attended by 90 per cent of the group’s members.

A total of 50 network engineers from 20 member companies used a specially designed lab to accomplish an equivalent of two months of quality assurance testing within a week, and resolved features and interoperability issues with different vendors, the organisation claimed.

ONF member companies had the opportunity to test new software features and products that had not yet been tested for interoperability with other vendors’ software and hardware products. OpenFlow versions 1.0, 1.2, and 1.3 were tested in commercial controllers and hardware and virtual switches, verifying implementations of OpenFlow-based SDN that would be commercially applied to service provider, datacentre, and enterprise markets.

OpenFlow 1.3 allows member companies to address complex network behaviour, optimise performance, and leverage a richer set of capabilities as well as extending to functions incorporating IPv6 and MPLS. IPv6 has become the standard in the datacentre, and testing showed the ability to add new resources and functions that make SDN and the OpenFlow protocol more scalable. In addition, the test cases for MPLS showed the movement of SDN and the OpenFlow protocol outside of the datacentre, with successful implementations performed to allow service providers greater control of the network with open and interoperable software.

“SDN is gaining wider acceptance in the service provider community as a transformative network concept, reinforcing the importance of continuing maturation and implementation of the OpenFlow protocol,” said Chris Janz, vice president of market development at Ciena. “In this latest ONF PlugFest, Ciena has collaborated with vendor and operator counter-parts in inter-op testing. After a week of testing, it is clear to us that there has been a significant maturing of the OpenFlow ecosystem,” he said.

Contributed by James Middleton, Managing Editor of telecoms.com

The Revolutionary Benefits of Telco SDN

Article contributed by Dimitris Mavrakis, Principal Analyst with Informa Telecoms & Media, who will be speaking in-depth about SDN and NFV at the forthcoming SDN World Conference in Barcelona on 11-13 June 2013, a uniquely operator-rich event featuring visionary contributions on the potential of SDN and NFV from operators around the globe.

With infrastrucDimitris Mavrakis smallture vendors offering their own flavors of SDN, the interest – and confusion – in the telecoms industry is increasing. Early adopters begin to deploy SDN concepts while mainstream operators attempt to understand the cultural, technical and business implications of the new concepts.

In the context of telecom networks, telco SDN refers to separation of the control and data layers, virtualization of network components and service exposure, although each vendor may have a different strategy according to their technical legacy and existing product lines.

In many cases, telco SDN reminds of previous, somewhat unsuccessful initiatives, including IMS. But in a way, SDN is already an established concept in the IT domain in the form of OpenFlow and OpenStack; although this is in a different context, it offers similar benefits to both IT and telecoms service providers. Increased scalability, flexibility, cost savings and less dependence on overprovisioning for network design are a few examples of short-term benefits SDN brings to both IT and telecoms. In the long term, the concept will allow for new service architectures, service exposure, new revenue opportunities and more efficient operations.

SDN and IMS promise – among other things – horizontalization of network infrastructure and easier service exposure to third parties but are without clear new business and revenue opportunities. However, there are several differences from previous technologies, especially IMS.

Telco SDN (and NFV) has serious advantages over IMS: SDN is already being implemented in the IT domain, and operators and vendors will have learned from their involvement in IMS. Also, SDN is attempting to enter the market by applying practices and technologies from the IT domain to the telecoms environment – this is in contrast with IMS, which was a completely new and very optimistic concept. Moreover, the vendors involved in SDN and virtualization also come from the IT domain and have experience of large-scale IP networks – which is the ultimate convergence of mobile networks.

What is likely to evolve first

The benefits of telco SDN increase towards the edge of the network and are maximum when facing the end user. For example, a virtualized CPE will not require continuous upgrades when a fixed operator upgrades to new technologies. On the other hand, the cost to virtualize the edge of the network – especially when considering radio networks – is prohibitively expensive.

LTE networks have just been deployed and operators will require some from of ROI from these existing networks before investing billions once again. Cloud RAN is the first implementation of virtualization in the radio access layer, where baseband processing is pooled at a semi-centralized location to allow for more fluid resource allocation. This also requires high expenditure and is likely to be deployed once small cells become widespread.

On the other hand, the Enhanced Packet Core (EPC) is a part of the network that is more likely to be virtualized in the short term, due to its flat (IP) nature and the need for components to be flexible and scalable. The core network is far more centralized compared to the RAN and as such, easier to upgrade, maintain and virtualize. EPC components include the MME (mobility management) and gateways (PGW, SGW) which may be implemented in software and run from data centres in order to provide a more scalable and flexible architecture. Specific functionality of the network may be exposed to trusted third parties while operators will be able to create customized services for groups of customers, or even individual premium customers (SMEs or enterprises).

Why Telco SDN is revolutionary

Despite the failure of previous network technologies, Telco SDN arrives at the market at the right time, when the telecoms and IT worlds are starting to align. Mobile operators will soon find that relying on traffic forecasts and overdimensioning may not provide an adequate business case for rolling out nationwide networks.

Despite the fragmented and multi-vendor nature of existing mobile networks, Telco SDN and similar concepts can provide a real answer to achieving more flexible, scalable and adaptive networks for real world scenarios.