Go to previous page / next page of Tier3 site log

28. 01. 2010 Test of data transfer latency at PSI

Method : subscribe one by one the following datasets:

1. /MinimumBias/BeamCommissioning09-BSCNOBEAMHALO-Dec19thSkim_341_v2/RAW-RECO
2. /MinimumBias/BeamCommissioning09-RecoTracks-Dec19thSkim_341_v2/RAW-RECO
3. /MinimumBias/BeamCommissioning09-SD_InterestingEvents-Dec19thSkim_341_v1/RAW-RECO
4. /MinimumBias/BeamCommissioning09-SD_InterestingEvents_MinBias10-Dec19thSkim_341_v1/RAW-RECO
5. /MinimumBias/BeamCommissioning09-SD_AllMinBias-Dec19thSkim_341_v1/RAW-RECO

and keep track of the latecy in the transfer begin, transfer speed and sources

Dataset	Size (GB)	Sources	Subscription	Transfer begin	Transfer Complete	Latency Time	Transfer Time	GB/h
1	190	FNAL, RWTH (few %)	2010-01-27 14:23:29	2010-01-27 15:23:56	2010-01-28 03:01:45	1:00	11:07	17
2	351	FNAL, RWTH (few %)	2010-01-28 13:34:13	2010-01-28 15:16:25	2010-01-29 17:10:36 (1)	1:45	~ 26h	13.5
3	288.35	RWTH	2010-02-01 22:14:37	2010-02-01 22:35:29	2010-02-02 02:11:25	20m	3:26	84
4	514.4	RWTH	2010-02-02 14:07:44	2010-02-02 14:33:01	2010-02-02 20:22:57	25m	5:50	88.1
5 (2)	2.5 TB	CSCS: RWTH / PSI:RWTH	2010-02-03 10:25:57	2010-02-03 11:08:17 / 2010-02-03 11:34:12 (from RWTH)	2010-02-03 13:32:47 / 2010-02-04 13:42:06	33m / 1:10h	2:24 /26h	815 / 96
TOTAL

(1) The transfer would have been completed around that time, but the last two files started failing at FNAL (timeout. Debug transfers from RWTH are ok, so a proxy issue excluded). Transfer actually ended ~31/1/2010, with the transfer of the last two files from RWTH

(2) Subscribed both at CSCS and PSI. As data is not flowing from CSCS, all sites but CSCS excluded from dowload agent at 2010-02-04 10:12. Added -ignore '*' at 2010-02-04 12:12:10. No actual effect as with the last modification all transfers were routed from RTWH and FNAL.

-- LeonardoSala - 2010-01-28

Go to previous page / next page of Tier3 site log