The KIP ATLAS computing cluster is an essential tool for the KIP ATLAS HEP group. It offers great flexibility and functionality with currently around 500 CPU cores, 400 Terabyte disk space and a 10 Gigabit fiber internet connection. The farm serves 3 main purposes:
- Hosting machines for daily interactive work
- Processing large amounts of data
- Storing large amounts of data
For day-to-day work, we provide 4 interactive servers with an operating environment identical to CERN machines (a custom CentOS 7). Multiple users can log in to these and process small amounts of data. Central CERN servers host the necessary ATLAS software for this, which is then cached locally. Our user’s home directories are stored in a central Raid10 array to provide the same environment on all machines. This is backed up daily both to URZ and locally on tape.
To process large amounts of data or perform computationally expensive offline analyses, the farm has 70 computing nodes. The High Throughput Computing Framework HTCondor ensures a fair share of these resources among our users.
Offline analyses require to download and store large amounts of data. That is facilitated by three Raid6 arrays with 400 TB capacity in total and a 10 Gigabit connection to our network of computing nodes. We provide an additional 24 TB Raid6 array for the output of these analyses.