Saved in:
| Main Authors: | , , |
|---|---|
| Format: | Preprint |
| Published: |
2026
|
| Subjects: | |
| Online Access: | https://arxiv.org/abs/2603.09568 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1866908906493575168 |
|---|---|
| author | da Silva, J M Costa, M A Iope, R L |
| author_facet | da Silva, J M Costa, M A Iope, R L |
| contents | This paper presents a detailed case study of the T2_BR_SPRACE storage frontend architecture and its observed performance in high-intensity data transfers. The architecture is composed of a heterogeneous cluster of XRootD [1] Virtual Machines (VMs) with 10 Gb/s and 40 Gb/s links, which aggregate data from a 77 Gb/s dCache [2] backend via pNFS to an external 100 Gb/s WAN link. We describe the system configuration, including the use of the BBR [3] congestion control algorithm and TCP extensions [4]. Under peak production conditions, we observed the system sustaining an aggregate throughput of 51.3 Gb/s. An analysis of a specific data flow to Fermilab (FNAL) showed peaks of 41.5 Gb/s, validated by external monitoring tools (CERN). This study documents the performance of a complex virtualized architecture under real load. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2603_09568 |
| institution | arXiv |
| publishDate | 2026 |
| record_format | arxiv |
| spellingShingle | Case Study: Performance Analysis of a Virtualized XRootD Frontend in Large-Scale WAN Transfers da Silva, J M Costa, M A Iope, R L Distributed, Parallel, and Cluster Computing This paper presents a detailed case study of the T2_BR_SPRACE storage frontend architecture and its observed performance in high-intensity data transfers. The architecture is composed of a heterogeneous cluster of XRootD [1] Virtual Machines (VMs) with 10 Gb/s and 40 Gb/s links, which aggregate data from a 77 Gb/s dCache [2] backend via pNFS to an external 100 Gb/s WAN link. We describe the system configuration, including the use of the BBR [3] congestion control algorithm and TCP extensions [4]. Under peak production conditions, we observed the system sustaining an aggregate throughput of 51.3 Gb/s. An analysis of a specific data flow to Fermilab (FNAL) showed peaks of 41.5 Gb/s, validated by external monitoring tools (CERN). This study documents the performance of a complex virtualized architecture under real load. |
| title | Case Study: Performance Analysis of a Virtualized XRootD Frontend in Large-Scale WAN Transfers |
| topic | Distributed, Parallel, and Cluster Computing |
| url | https://arxiv.org/abs/2603.09568 |