Ethernet for High-Throughput Computing at CERN

Krawczyk, Rafał Dominik; Colombo, T.; Neufeld, N.; Pisani, F.; Valat, S.

doi:10.1109/tpds.2022.3163472

Cited by 2 publications

(2 citation statements)

References 24 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The estimated minimum required buffer sizes to avoid packet drops on the QFX switch are shown in Figures 4 and 5 as a function of the event size and event rate respectively. Experiments cover event rates in the range [1][2][3][4][5][6][7][8][9][10][11][12][13][14][15][16][17][18][19][20] kHz and event sizes in the range [1][2][3][4][5][6][7][8] MB, but fewer cases are included in the plot for simplicity. Regarding lab measurements, the plots show that the system can not cope with high event rates or event sizes due to lack of buffer.…”

Section: Resultsmentioning

confidence: 99%

“…In a recent performance study conducted within the context of the LHCb experiment and utilising Ethernet RocEv2 (instead of TCP as in this study), a comparison between deep and shallow buffer switches was performed. The study revealed that the shallow buffer switch exhibited sub-optimal performance, while the deep-buffer switch demonstrated satisfactory performance but only when dealing with a limited number of data sources [14].…”

Section: Related Workmentioning

confidence: 99%

See 1 more Smart Citation

Benchmarking Data Acquisition event building network performance for the ATLAS HL-LHC upgrade

Pozo Astigarraga,

Bonaventura,

Maple

et al. 2024

EPJ Web of Conf.

View full text Add to dashboard Cite

The ATLAS experiment’s data acquisition (DAQ) system will be extensively updated to take full advantage of the High-Luminosity LHC (HL-LHC) upgrade, allowing it to record data at unprecedented rates. The detector will be read out at 1 MHz, generating over 5 TB/s of data. This design poses significant challenges for the Ethernet-based network, which will have to transport 20 times more data than during Run 3. The increased data rate, data sizes and number of servers will exacerbate the TCP Incast effect observed in the past, making it impossible to fully exploit the capabilities of the network and limiting the performance of the processing farm. We present exhaustive and systematic experiments to define buffering requirements in network equipment to minimise the effects of TCP Incast and reduce the impact on processing applications. Both deep and shallow buffer switches were stress-tested using DAQ traffic patterns in a test environment at approximately 10% of the expected HL-LHC DAQ system size. As the desired HL-LHC system hardware is not currently available and the laboratory size is significantly smaller, the tests aim to extrapolate buffer requirements to the expected operating point. A novel analytical formula and new simulation models have been developed to cross-validate the results. The results of these evaluations will contribute to the decision-making process for the acquisitions of network hardware for the HL-LHC DAQ.

show abstract

Section: Resultsmentioning

confidence: 99%

Section: Related Workmentioning

confidence: 99%

Benchmarking Data Acquisition event building network performance for the ATLAS HL-LHC upgrade

Pozo Astigarraga,

Bonaventura,

Maple

et al. 2024

EPJ Web of Conf.

View full text Add to dashboard Cite

show abstract

Machine Learning Approach Towards Telemarketing Estimation

Saltı,

Kangal,

Zengin

2024

Middle East Journal of Science

View full text Add to dashboard Cite

Machine learning (ML) provides us the flair to achieve knowledge from large data pools that we as human beings may not study. So, the ML is a process in which we train computers to recognize patterns in a dataset, such that they learn that significant information and use them to achieve new things that they haven’t encountered before. In this study, an attention has been paid on investigating whether successful predictions can be achieved for a company's decision- support mechanism by analyzing a selected telemarketing dataset in the ML algorithms. Thus, a remarkable conclusion has emerged from our investigation: the “Age” of customer and “Product ID” are the most important attributes that affect the number of “Sales”.

show abstract

Ethernet for High-Throughput Computing at CERN

Cited by 2 publications

References 24 publications

Benchmarking Data Acquisition event building network performance for the ATLAS HL-LHC upgrade

Benchmarking Data Acquisition event building network performance for the ATLAS HL-LHC upgrade

Machine Learning Approach Towards Telemarketing Estimation

Contact Info

Product

Resources

About