Clustering-preserving Network Flow Sketching

Fu, Yongquan; Li, Dongsheng; Shen, Siqi; Zhang, Yiming; Chen, Kai

doi:10.1109/infocom41043.2020.9155388

Cited by 11 publications

(13 citation statements)

References 23 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…OmniMon [20] seeks full accuracy and resource efficiency over collaborated network entities. LSS [14] applies a cluster-preserving approach to reduce the estimation error, by guaranteeing that each network flow is always mapped to the closest cluster. Thus, to process packet streams like traditional synopsis, LSS maintains an in-memory cache for active network flows and dynamically adjusts the clusters for these network flows.…”

Section: B Related Workmentioning

confidence: 99%

“…1) Per-flow Query: We first test the relative errors of the per-flow query. We compare Jellyfish with five per-flow sketching methods including count-sketch (CS) [16], cusketch (CU) [1], count-min (CM) [15], Elastic Sketch (ES) [12] and LSS [14].…”

Section: B Comparisonmentioning

confidence: 99%

“…2) Entropy Query: We next compare the relative errors of the entropy query of Jellyfish with five entropy-sketching methods including count-min (CM) [15], count-sketch (CS) [16], Sieving [27], Elastic Sketch (ES) [12] and LSS [14].…”

Section: B Comparisonmentioning

confidence: 99%

“…3) Heavy Hitter: We next compare the F1 scores of the heavy-hitter query with six heavy-hitter methods including count-min (CM), count-sketch (CS), Spacesaving (SS) [2], ElasticSketch (ES), and hashpipe [4] and LSS [14]. We set the threshold of the heavy hitters to the top-5% of network flows.…”

Section: B Comparisonmentioning

confidence: 99%

“…Several approaches [12], [13] proposed to separate large ones from the rest to reduce the peak estimation error for the rest. LSS [14] proposed to map similar network counters to the same bucket array in order to reduce the approximation error. Unfortunately, LSS needs to keep all active network flows in the memory, in order to update the mapping of network flows to the nearest clusters in real time as the network flow counters keep increasing, which is challenging with respect to large-scale network traffic.…”

Section: Introductionmentioning

confidence: 99%

See 4 more Smart Citations

Jellyfish: Locality-Sensitive Subflow Sketching

Shen

et al. 2021

IEEE INFOCOM 2021 - IEEE Conference on Computer Communications

Self Cite

View full text Add to dashboard Cite

To cope with increasing network rates and massive traffic volumes, sketch-based methods have been extensively studied to trade accuracy for memory scalability and storage cost. However, sketches are sensitive to hash collisions due to skewed keys in real world environment, and need complicated performance control for line-rate packet streams.We present Jellyfish, a locality-sensitive sketching framework to address these issues. Jellyfish goes beyond network flow-based sketching towards fragments of network flows called subflows. First, Jellyfish splits consecutive packets from each network flow to subflow records, which not only reduces the rate contention but also provides intermediate subflow representations in form of truncated counters. Next, Jellyfish maps similar subflow records to the same bucket array and merges those from the same network flow to reconstruct the network-flow level counters. Realworld trace-driven experiments show that Jellyfish reduces the average estimation errors by up to six orders of magnitude for per-flow queries, by six orders of magnitude for entropy queries, and up to ten times for heavy-hitter queries.

show abstract

Section: B Related Workmentioning

confidence: 99%