An Innovative Network Management Based Solution for Managing Hybrid Networks for Large

An innovative Network Management Based Solution for managing Hybrid Networks for Large Scale Events

MICHALIS ELLINAS, LAMPROS RAPTIS, NIKOS DRAGIOS, PANAGIOTIS TRAKADAS

Department of Electrical Engineering and Computer Science

National Technical University of Athens

9, Heroon Polytechniou, 15773 Zographou, Athens,

GREECE

THEODORE ZAHARIADIS, KOSTAS VAXEVANAKIS

Ellemedia Technologies Ltd

223 Syggrou Av., GR-17121, Athens

GREECE

STAMATIS VOLIOTIS, CHRISTOS MANASIS

Technological Educational Institute of Chalkida

Psahna, GR-34100, Evia

GREECE

Abstract: - The coverage of large scales event such as the major athletic or cultural events, requires the deployment of a complex network architecture capable of fulfilling the tremendous bandwidth requirements of such scale networks with guaranteed Quality of Service (QoS) to the various users (residential, corporate, ISPs, TV channels, etc) of such network. The purpose of this paper is to propose an innovative network management based solution for the management of an hybrid IP over WDM network, which consists the core of an extended Content Distribution Network (CDN). The CDN utilizes the features provided by the proposed Integrated NMS (INMS) to establish, monitor and preserve content distribution, while at the same time implements rudimental traffic engineering procedures to avoid network congestion and service failure. The supported bandwidth, the ability of the NMS mainly to guard against performance degradation and finally the expected service provisioning time window indicate the suitability of the proposed architecture for the coverage of large scale events.

Key-Words: - Network Management, Content Distribution Networks, Integrated Management, Optical Networks

1  Introduction

It is common ground that the worldwide experienced expansion of the Internet has been unprecedented. This phenomenon is expected to continue in the years to come following two vertical axes, the first being the regional coverage and the second and most important the great increase of the request for bandwidth accompanied by certain and often acute performance restrictions, driven by the evolvement of the network to offer more and mostly QoS-aware services. In fact, the biggest challenge the Internet community is called to meet is the adaptation, enhancement and reordering of the infrastructure, the network and finally the established procedures in order to cope with the new and expanding role Internet plays in nowadays communications. In summary, the demand from both corporate and individual users for QoS support poses a major financial opportunity that stands behind the rush of operators for their network expansion and optimum resource utilization and the emergence of new business players, bandwidth and service brokers.

This paper is intended to describe a network architecture, focusing on large-scale events such as big athletic or cultural events, that combines the realm of network management with content distribution. The structure of the paper is the following: section 2 initially presents the network architecture needed for the efficient deliver of content to different users focusing on the core of such a network. Then, the major requirements due to the nature of such event are presented, followed by different scenarios. The concept of CDN is further explained and the way that a CDN can exploit the proposed NMS solution is also analyzed. After the discussion of the overall network architecture, the proposed NMS solution is explained, concluding with the major results as well as future expansion of our work.

2  Network Architecture

2.1  Abstract network architecture

Figure 1 depicts a simplified view of the targeted network architecture. The key elements include business roles, technology – specific sections of the network and physical (athletic or not) premises. The network segments can be categorized according to a number of criteria. With respect to the proposed Network Management based approach, the network architecture is differentiated to the NMS controlled, including the Core Network and a number of surrounding routers or even IP networks and the NMS independent sections.

Fig.1: Abstract Network Architecture

The proposed NMS provides certain IP Points-of-Presence (IP PoP), which are advertised to all potential users. These IP PoPs are fixed and the dissemination of the relevant information is carried out by means outside the scope of the NMS itself. The communication of all interested parties with the NMS is based on the CaSMIM interface. The concept follows a client – server schema with the NMS functioning as the server that receives requests for the provisioning of connections between any of its declared PoP. The request involves the relevant end point as well as the expected QoS parameters. The resulting connection can be abstracted as a direct link between the parties involved. The NMS is accountable for monitoring the status of the network resources and that of the established links. The clients are kept informed for the status of their requests and the respective link status. This is the only piece of information the NMS disseminates.

2.2  The need for scheduling

The ideal solution would have been an optical backbone network that utilizes control plane functionality. The integration of the optical backbone network control plane with the existing in the “IP-world” functionality would have among other things led to the automation in the provisioning procedures and the shortening of the amount of time needed for a connection to be established. Although significant effort is been placed towards the specification – implementation of an optical control plane and the interaction with the IP control plane, the goal has not been accomplished. The proposed Network Management level approach is a midterm solution. It does, however, pose certain limitations, most importantly: the provisioning time window lies in the order of minutes. Since real – time coverage requirements urge for provisioning procedures that conclude in no more than a few seconds, a significant predicament seems to form.

On the other hand, one should bear in mind the particularities of the event the system attempts to cover. For example, the Olympic Games is a highly ordered – scheduled chain of events, which are performed in place and in time according to a schedule that has been published long before the launching event.

Therefore, a scheduler at the highest level could be used to overcome lengthy provisioning time window obstacle. That would imply that the potential users of the system should inform the system administration about the required services prior to the launching of the respective event. The amount of time could not be less that the provisioning time window, a typical value of 5 minutes.

2.3  Service Consistency Guaranties

Consistency in service provisioning is among the higher priorities in live (near real time) coverage. In the case of a major event of global interest the importance of uninterrupted and unimpeded content distribution can not be exaggerated. The functionality provided by the Network Management System that is responsible for the efficient and persistent operation of the backbone (core) network constitutes the cornerstone towards this end. Although the Network Management System will be discussed in details in a following section, the relevant requirements can be stated in advance.

With regard to the core network, we could differentiate the pertinent procedures into two major areas:

1.  Procedures and functionality that monitor traffic load and network performance, guarding against burdensome network conditions. These correspond to the Performance Management (PM) Functional Area of the Network Management System.

2.  Procedures and functionality that assume control in cases where either PM functionality reports that the traffic load conditions have crossed certain thresholds, or some network element has come out of operation. Subsequent actions involve link – service restoration based on the corresponding service priority in the Service Level Agreement (SLA). The procedure of this area correspond to the Fault Management (FM) Functional Area of the Network Management System.

The NMS dedicated section that follows illuminate the issue with further details.

2.4  Scenarios

The envisaged system should be able of delivering high bandwidth along with ensuring QoS constrains. The anticipated usage of the proposed architecture can be largely described as a (Near) Video on Demand service required by: ISPs, content distributors, or other interested parties.

The prerequisite and key element for a successful implementation of such a system is the unimpeded cooperation between the building blocks, i.e. the NMS and the operation centers of each client. Baring in mind that in the general case these systems are administrated by individual market players, the outcome is bound by the following factors:

1.  Mechanisms and procedures for adequate and, mostly, safe information sharing between the systems involved.

2.  Mechanisms and procedures that will allow the formation of new and monitoring of established high-level SLAs. The SLAs should clarify a flexible accounting schema and guard against information and resource misuse.

2.5  Content Distribution Network

2.5.1  CDN Concept

The idea behind Content Distribution Network technologies is to place content at servers close to network edges and redirect users requests to those servers, instead of letting users request rich content from its origin location. Given this, CDN is an independent network of dedicated servers, we call them surrogates, that content providers can use to distribute their contents throughout the Internet [1]. CDN contains contents from multiple content servers and enables quick and reliable response to clients. A request may be served from a surrogate containing the requested object from a closer to client location than the remote content provider. Fig.2 shows the topology of a CDN. At the network edges there are the content providers that publish their content to the CDN. Surrogates are also at the edges of the backbone network, closer to clients’ access network, in order to serve clients on behalf of the origin content providers.

Fig.2: CDN Architecture

CDNs are on the way to alleviate the long delays experienced by users and deliver high quality rich media content via Internet in affordable time. CDN facilities are between the end user and the content provider. They can be applied in several forms either by the content providers or the ISPs, through which users have access to the Internet. CDN reduces content providers’ network traffic load, improves performance of the origin servers, as processing time at the origin site is decreased. Consistency and availability of content are the main issues content owners spend their time and money for since the delivery of content is a CDN’s task. Even more, Internet traffic load is reduced since servers with replicated content are now closer to the end user and requests for content are limited to close geographical areas. As a result clients do experience less response delays even for rich content and bandwidth demanding applications.

The evolution of new demanding services and the increasing number of Internet users makes Internet slower. Although bandwidth improvements have been done with the use of fiber optic pipes and high-speed last mile technologies, problems of access delay exist due to the richness of today Internet content. Data will continue to be richer resulting in the capacity of servers and bandwidth of the Internet infrastructure to be insufficient. Especially for streaming media applications the problem of extended delays experienced by users is a reality. Streaming enables audio, video and animation files to be sent and viewed during the delivery process so that the user can enjoy the content before the whole transmission comes to an end.

What has to be done, in order to exploit the CDN technologies to deliver streaming media with affordable time delays, is to ensure consistency and reliability for the web site visited by the end users. Servers of CDN existing at network edges and replicating static content should be able to guarantee some QoS even for content that has to be sent on demand. The quality of such content varies widely and depends basically on the type of the media delivered and the speed of the pipe between the origin and the replica server. If we could allocate and guarantee for the QoS over the channel between those later end points, we would actually achieve the needed quality for streaming media delivered to client. NMS service can do that for the CDN providers. Based on the user’s request for a certain level of QoS and the application’s bandwidth requirements NMS will establish the best tunnel for the transmission of the streaming media requested. Besides, monitoring and fault management mechanisms assure for the maintenance of the network conditions over the channel established.

2.5.2  CDN requirements

Apart from the procedures that a CDN normally implements, the content distribution network operation center should exploit the functionality that the NMS provides. Under this perspective, the following requirements can be stated:

1.  In the case that the NMS established connections will be used by more than one CDN-specific flows, trunking procedures lie under CDN responsibility

2.  CDN operation center should invoke rudimental traffic engineering techniques, in order to conserve resources and minimize costs.

3.  CDN operation center should assign priorities for the NMS established connections. A simple schema could consist of assigning high priority to real or near real time services, leaving lower priority connections to less demanding procedures such as mirroring and cashing.

2.5.3  The CDN scenario

Concluding this section, the following Fig.3 briefly depicts what the situation would be like in the case of a CDN.

Fig.3: The CDN scenario

Note the CaSMIM based communication between operation centers, and the abstraction of the newly INMS created link (red dotted line).

3  The Network Management System

3.1  Optical – Electrical Interconnection

The constantly increasing demand for higher bandwidth, the need to narrow service provisioning time window, the necessity for robust and resilient networks and finally the acknowledgement that IP traffic tends to become the dominant constituent in tomorrow’s communications market gave rise to the world research community to investigate and evolve procedures that would:

1.  Offer the connectionless IP some useful characteristics that are inherent to connection oriented protocols.

2.  Utilize the huge amount of bandwidth fiber optics provides and shift this potential towards the IP perspective.