QoS on Data Centers

Many years ago, network engineers design one network for each service. That provides big, complex and costly networks. For that, the trend was to integrate services in only one network, to notably simplify it in terms of cost and management. But to assure the correct functioning on critical services in huge load cases, is needed to implement Quality of Service (QoS) which permits manage traffic depending on its classification. In this way, we are reducing all networks in one, prioritizing traffic and shape the traffic to attend critical traffic such as Voice over IP, videostreaming or other real time traffic.

To say in an academic mode, Quality of Service is a set of technologies for managing network traffic in a cost-effective manner to enhance user experiences for home and enterprise environments. QoS technologies allow to measure bandwidth, detect changing network conditions (such as congestion or availability of bandwidth), and prioritize or throttle traffic. QoS technologies can be applied to prioritize traffic for latency-sensitive applications (like voice or video) and to control the impact of latency-insensitive traffic.

We have some indicators which permits characterize Q0S on our network: bandwidth, end-to-end delay, jitter and packet loss. Monitoring these parameters for each traffic type we can assure availability on some of these types.

To classify traffic, we can do through protocol tools which support QoS, mainly based on IP protocols, which provides the Type of Service (ToS) field or DSCP field where we can classify traffic in 8 types or 64 types respectively.

Other protocols which include QoS capabilities are Ethernet, FiberChannel, MPLS or ATM.

It’s important to apply QoS on a Data Center to assure critical services and give them the priority that they require at the moment required, and other less important services could give these resources to critical services. However, applying QoS will increase the design and management complexity.