作者: Emine Ugur Kaynar , Mania Abdi , Mohammad Hossein Hajkazemi , Ata Turk , Raja R. Sambasivan
DOI: 10.1109/BIGDATA47090.2019.9006396
关键词: Network congestion 、 Bandwidth (computing) 、 Server 、 Computer science 、 Computer network 、 Upstream (networking) 、 Default gateway 、 Cache 、 Network topology
摘要: Current caching methods for improving the performance of big-data jobs assume high (e.g., full bi-section) bandwidth; however many enterprise data centers and co-location facilities have large network imbalances due to over-subscription incremental networking upgrades. We describe D3N, a multi-layer cooperative architecture that mitigates by on access side each layer hierarchical topology, adaptively adjusting cache sizes based observed workload patterns congestion. added (and submitted upstream) 2-layer D3N Ceph RADOS Gateway; read bandwidth achieves 5GB/s speed our SSDs, we show it substantially improves job while reducing traffic.