Problem solved by the authors
Authors are trying to solve congestion issues encountered in data centers. It is difficult to tune and monitor networks if they take about one second to react to flows (Rasley et al n.p ).
Relevance of the problem
The challenge addressed by the author is crucial because it increases efficiency and effectiveness in data centers (Rasley et al n.p ). In addition, such problems make data systems to operate at scales and speeds that allow the operators to react to rapidly to transient problems such as congestion.
The central idea
In the paper, the author is explaining how self-tuning networks can be utilized for monitoring network conditions. Notably, some systems can react immediately to events like congestion. However, conventional systems take about one second to link utilization of flows (Rasley et al n.p ). The central idea described in the paper is how to come up with software that can react to congestions in a millisecond.
How the work differs from the previous
Planck network outlined in the paper differs from the other networks in that it takes the shortest time to respond to issues. Furthermore, the measurement architecture employed in Planck system uses port mirroring to acquire information at 280 milliseconds to 7 microseconds, unlike the previous networks that took about one second to detect congestions (Rasley et al n.p ). In short, Planck network has high speed and accuracy, it can detect and reroute flows in milliseconds.
Major flaws in the work
Planck systems are associated with numerous flaws. The fact that all micro-benchmarks are controlled by one switch hampers the detection and rerouting of flows in the network. Similarly, the switches used in Planck networks do not accommodate port congestion. Therefore, modern switches can reduce the flows associated with the novel software. Sample latency cannot be measured precisely in Planks because of some delays. To minimize technical sample delays, NIC applications are used. The applications provide measurements that can accurately estimate the sample latency (Rasley et al n.p ). However, Planck allows applications such as SDN that enable the network to detect events within milliseconds.
In conclusion, the paper has numerous ideas on how to develop software-defined networks that will be needed in 10 years. Ideally, through careful engineering process Planck network can lead to a much better system that can react rapidly to congestions.
Works Cited
Rasley, Jeff, Brent Stephens, Colin Dixon, Eric Rosner, and Wes Felter. Planck: Millisecond-scale Monitoring and Control for Commodity Networks (n.d.): n. pag. Web.
