Quality Assurance

With networks handling increasing amounts of diverse traffic, Quality of Service is emerging as a critical tool in the timely delivery of business critical applications and data

  • E-Mail
By  Zoe Moleshead Published  December 4, 2002

Quality of Service|~||~||~|As with every aspect of the IT infrastructure, networks are continuing to expand rapidly, as they handle increasing amounts of voice, data and video traffic. As users look to capitalise on the cost savings and efficiencies that converged networks can offer, they are also placing increasing emphasis on the prioritisation of traffic. With this demand has come the emergence of Quality of Service (QoS).

The role QoS plays in the network infrastructure spans beyond just filtering different types of traffic. It is also critical to ensuring the timely delivery of core business information and applications to network users.

“Contrary to what many people believe, voice is not the driving factor behind Quality of Service. Of course voice requires QoS, any converged network that runs both voice, data and video requires QoS. However, the biggest push is from enterprises that are running mission critical applications on their enterprise network and want to make sure that these applications get the quality they require,” explains Anwar Kotob, systems engineering manager, Cisco Systems, UAE.

With increasing amounts of data, particularly graphically bandwidth intensive information, being pushed over enterprise networks, bottlenecks are not uncommon. As such, enterprises are becoming increasingly dependant on QoS to prioritise their business critical applications.

“Enterprises need QoS to ensure that their mission critical applications, those that will make a difference to their bottom line, operate independent of any other support applications that are running on the network,” says Kotob.

While prioritising business critical traffic is one element of QoS, both enterprises and service providers face the challenge of balancing voice, data and video services. And, with this, comes an additional set of criteria.

“Services like voice and video are real-time services,” says Emad Makiya, general manager, Extreme Networks, Middle East & North Africa. “[These] differ from data services, in that they have very strict requirements on delay, latency and jitter. Only with the use of the correct QoS policies can these requirements be fulfilled,” he adds.

These layers of traffic can also be broken down further, with video services encompassing everything from videoconferencing, live TV broadcasts and distance learning, while voice extends to voice over IP (VoIP).

And while users are increasingly migrating to IP-based networks to handle their voice, video and data traffic in one centralised and simplified infrastructure, this also brings with it some challenges for Quality of Service.

“With the explosive growth of the internet, most network traffic today is IP-based. Having a single end-to-end transport protocol is beneficial because networking equipment becomes less complex to maintain. This results in lower operational costs. This benefit, however, is countered by the fact that IP is a connectionless protocol — IP packets do not take a specific path as they traverse the network — this results in unpredictable QoS,” says Ahmad Antar, senior systems engineer, Nortel Networks.

When it comes to implementing QoS, the process is carried using classification or Class of Service (CoS), queues and queue servicing algorithms that deal with separate traffic flows.

On a basic level this means traffic is prioritised and classified — in line with the designated QoS parameters — at the edge of the network, and this classification is then carried through into the core network.

“As the traffic traverses the network devices within the backbone it will have to obey the classification and deal with it — when the core gets high priority packets it must prioritise that traffic over the normal or low priority packets,” says Kotob.

At a more complex level, QoS also enables users to reduce levels of latency, jitter and loss by allocating specfic bandwidth and prioritising traffic packets.

“For example, real-time voice traffic can be guaranteed with a specific amount of bandwidth and a certain priority to minimise packet-drop, latency and jitter. Data traffic is specified with a lower priority, while H.323 call signalling information is assigned an even higher priority to ensure quick call set-up times,” explains Extreme’s Makiya.

||**||Implementing QoS|~||~||~|Vendors, such as Extreme, Foundry, Nortel and Cisco provide QoS functionality in their network switches and routers, to ease the management and configuration tasks for network administrators. It is, however, still important for network engineers to monitor and measure whether QoS is meeting their designated requirements.

At an enterprise level this enables users to make sure they are not exceeding their network capabilities and provides a clearer picture of the impact QoS is having on their network as a whole.

“Sometimes QoS can inadvertently deprioritise a particular protocol and all these protocols make a difference to the network so users should measure them to see that the network is meeting the required QoS behaviour,” advises Kotob.

“Enterprises have an advantage in being able to track all their configurations. They have a single networking department that will configure all their edge and core devices. Whatever configuration they decide at the edge of the network, they can implicitly trust,” he adds.

At a service provider level, QoS has a larger significance because it is tied into customer service levels. If service providers are to fulfil the standards specified in a service level agreement (SLA), it is essential for both service provider and customer to be able to monitor the QoS levels they are giving or receiving.

Customers can determine the QoS they are receiving by measuring parameter including network availability, bandwidth, delay, jitter and loss.

“Service providers’ bandwidth is a lot more costly and they have to use that bandwidth a lot more intelligently,” comment Kotob.

“Furthermore, since the QoS performance may be specified in a SLA the service provider needs to ensure that the network is providing the performance as specified.

Once a service provider can accurately measure the network capabilities and provide a guaranteed performance level, they can confidently offer a billable service to subscribers,” confirms Antar.

Service providers also face challenges when trying to deploy Quality of Service against the internet. The global nature of the web means that traffic frequently traverses more than one service provider network. As such, QoS has to be agreed between all these carriers for it to be implemented effectively.

“When the internet was built, it was not built with QoS in mind because it is an international network,” confirms Kotob.

“A service provider serves a certain geography or space and they will have to talk to other carriers to get the traffic from point A to point B. QoS is a network-wide issue, so all those carriers along the way have to agree on these QoS mechanisms and contracts. This makes it a lot more difficult to establish QoS on the internet,” he continues.

Enterprises benefit from the developments of vendors and the inherent QoS functionality present in their products. At the service provider level, multi-protocol label switching (MPLS) is beginning to gather momentum, as users look to apply QoS more convincingly in their core networks. And with network traffic continuing to grow, QoS is set to become increasingly important.||**||

Add a Comment

Your display name This field is mandatory

Your e-mail address This field is mandatory (Your e-mail address won't be published)

Security code