QoS in our Data Center

In this second post of the day we are going to talk about how are we going to apply QoS techniques in the case study's data center. As it was discussed on the previous post a complex QoS setup is not always the best solution to actually guarantee the QoS requirements of the network. In the concrete case of a banking related data center we will need to apply some specific QoS techniques in order to lower the latencies of sensitive traffic as much as possible although those will be very specific applications. In a general sense we'll mostly focus on guaranteeing that there is no congestion on the network and that the throughput at all its points is the required for the actual demand and QoS requirements. Hereby, we'll oversize the bandwidth on the critical points to make it harder for elephant traffic generated delay to even happen.  Also congestion detection mechanism and most of all ECN Marking will be applied. The last one allows to detect congestion and reduce transmission rate before packets get dropped by the network hence improving the network performance by preventing retransmissions. As it's been said, by being in a banking environment, it will mostly have some really sensitive applications as stock trading ones which require a special treatment hence requiring the implementation of QoS mechanisms "per se". In those cases, it is important to keep the network configuration as clean as possible. Police ingress traffic would queue packets based on DSCP or 802.11p markings. Application aware processing which is appearing a lot in recent software for network devices would be best implemented on hypervisors or end hosts and not in the ToR switch.