Software Development for Openflow Switch Based on Network Processor of Mellanox Company

Smelyanski, R. L.; Васин, В.В.

doi:10.14489/vkit.2017.02.pp.031-039

Cited by 2 publications

(3 citation statements)

References 0 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…In order to register workloads and communicate with the controller, we add Saba API to the Spark and Flink frameworks. To do so, we modified the existing job submission and RDMA-enabled shuffle manager [43] in Spark to invoke Saba's interface functions. The individual workloads required no modification to support Saba.…”

Section: Saba Librarymentioning

confidence: 99%

Saba: Rethinking Datacenter Network Allocation from Application's Perspective

Katebzadeh

Costa

Grot

2023

Proceedings of the Eighteenth European Conference on Computer Systems

View full text Add to dashboard Cite

Today's datacenter workloads increasingly comprise distributed data-intensive applications, including data analytics, graph processing, and machine-learning training. These applications are bandwidth-hungry and often congest the datacenter network, resulting in poor network performance, which hurts application completion time. Efforts made to address this problem generally aim to achieve max-min fairness at the flow or application level. We observe that splitting the bandwidth equally among workloads is sub-optimal for aggregate application-level performance because various workloads exhibit different sensitivity to network bandwidth: for some workloads, even a small reduction in the available bandwidth yields a significant increase in completion time; for others, the completion time is largely insensitive to the available bandwidth.Building on this insight, we propose Saba, an applicationaware bandwidth allocation framework that distributes network bandwidth based on application-level sensitivity. Saba combines ahead-of-time application profiling to determine bandwidth sensitivity with runtime bandwidth allocation using lightweight software support with no modifications to network hardware or protocols. Experiments with a 32server hardware testbed show that Saba improves average completion time by 1.88× (and by 1.27× in a simulated 1,944server cluster).

show abstract

Section: Saba Librarymentioning

confidence: 99%

Saba: Rethinking Datacenter Network Allocation from Application's Perspective

Katebzadeh

Costa

Grot

2023

Proceedings of the Eighteenth European Conference on Computer Systems

View full text Add to dashboard Cite

show abstract

“…K6 in Figure 3 indicates where the RTP payload is extracted to identify the type of sub-flow, and K10 and K11 indicate where the video quality and temporal layers IDs are extracted to know what quality and temporal layers a video flow are associated to, respectively. These video layer meta-data are extracted from the Scalable H.265 video codec [47], [48], which is the latest video codec, to determine which layer can be removed to reduce the resolution or the frame rate of a given video flow to be able to adapt the video in network congestion conditions by removing less important video layers (more details in section 5). One of the challenges associated to the classification of the RTP is that DPI needs to be performed to identify correctly that 370 the payload is an RTP payload.…”

Section: Parsing Stagementioning

confidence: 99%

“…The first aspect to analyze is the maximum throughput that can be achieved using the standard Gtek Mellanox ConnectX-2 MLX-6801-1S 10GbE NIC to send and receive PCAPs from the NetFPGA card with the latest version of the Mellanox Linux drivers provided for the card, v.4.2-1.0.1.0 [48]. To do so, an iperf stress test has been carried out using a loopback interface and tcpdump to store both sent and received PCAPs.…”

Section: Performance Testmentioning

confidence: 99%