With our tape site properly opened now for about a year it is interesting to see how we are doing. The plot of the data ingestion at MIT throughout the last year tells the story. Initially, in November/December of 2024, there were some tests on various channels which eventually resulted in MIT being declared open for production writing by the end of January.
The file transfer servers (FTS) of the experiment started to write basically at 10 Gbps speeds which we thought was odd, because we were hoping for rates more consistent with 100 Gbps. It turned out that MIT is only connected with a 10 Gbps line to MGHPCC at Holyoke and we were taking over the complete line and thus interfering with other activities in the MIT installations at the MGHPCC. the solution was to throttle CMS transfers to make sure everybody could work.
Since then (March), the CMS collaboration has been pushing data into the MIT storage successfully with a few blips due to software upgrades etc. Recently in October, network upgrades has been performed and our rate has gone up significantly again. The MIT site has received 10.1 PB of production output so far. We are promised by the networking group that MIT will be connected at least at the 100 Gbps by the end of January and later eventually grow to 400 Gbps.
The output of the MIT tape system is also interesting of course. The transfers works well but with no major reprocessing ongoing and most datasets being on disk the activity it limited.
