the review of “Data Center TCP”
3010216110 杨典
How to came up with the idea?
Data Center requires highly available,highly puting and storage infrastructure using low ponents. The way of using novel architectures to build up better Data Center with low-cost es to a challenge.
In the paper, the author tests the improvement of DCTCP protocol by applying soft real-time applications as an example. Because those applications have same features such as low latency for short flows, high burst tolerance, and high utilization for long flows. First two requirements derived from Partition/Aggregate workflow pattern, once not been properly solved, would directly impact the quality of the result and revenue. The third demand, comes from continuously update internal data structures of these applications, is the essential of low latency and burst tolerance.
What can be solved?
The author made to contributions in the paper. First, they measure and analyze production traffic, extracting application pattern and needs. Second, they came up with Data Center TCP, by using a feature called Explicit Congestion Notification, effecting a novel control scheme at the sources. It extracts multibit feedback on congestion in work from single bit stream of ECN marks and use that estimate as a signal for the extent of congestion to operate with low buffer occupancies while achieving high throughput. More
网络论文 来自淘豆网m.daumloan.com转载请标明出处.
