The BRC has invested in a High Performance Computing system, enabling researchers to more quickly analyse, store and archive vast quantities of data generated by our sequencing applications.
Large BRC Cluster (Athena)
The large BRC HPC Cluster comprises some 33 servers for computation provided by 16x IBM iDataplex 2U chassis containing 32x IBM IDataplex nodes each fitted with:
-
2x 2.66Ghz Intel Nehalem processors (therefore 8 cores per node and a total of 256 processor cores for the cluster excluding heads nodes and the large node)
-
48GB memory
-
250Gb hard drive
A larger compute node an IBM x3755 server fitted with:
This gives a total of 280 CPUs for general cluster applications. Two master nodes IBM x3650 M@ servers with 2x Intel Nehalem 2.26GHz quad core processors with 24 GB RAM provide redundancy control and provisioning of the Cluster with the Sun Grid Engine for job scheduling and IBM xCAT for general cluster maintenance.
A key feature of the cluster is the data storage which uses Panasas ActiveStor Series 8 clustered storage using its built-in Panasas ActiveScale distributed parallel file system which enables much faster read and write access to the data using a high-performance storage system, rather than waiting for data to be extracted from a traditional, slower storage system. The current total capacity of the storage is 180TB with a usable amount of 146TB.
The cluster is housed in two 42U IBM racks in the GSTT NHS Data Centre and protected by UPS and generator supplies from power outages. The Cluster is backed up via the network to tape via a separate backup server to an IBM TS3310 Tape Library Unit with Tivoli Storage Manager housing up to 200 tapes in the library to enable long-term, secure (data encrypted), off-site data back-up.
Small BRC Cluster (Hera)
The small BRC HPC Cluster comprises some 9 servers for computation with:
-
4x HP G6 x86_64 compute nodes (2.6GHZ, 28GB RAM) running
-
4 Tesla S1070 Nvidia GPUs for novel GPU processing applications
-
3x Supermicro compute nodes (2.6GHz, 32GB RAM)
The above are combined to provide 104 CPUs for general cluster applications. The system is supported by 22TB of disk space in a RAID 6 array on a local HP Storage server comprising 2 MSA60 storage enclosures and all managed through a single head node (2.4 GHz, 16GB RAM).
The cluster is managed using the Rocks+ Cuda Rolls operating system. The cluster is housed in a single 42U APC rack located on the 7th Floor Genomics Lab and protected by APC UPS from power outages. The servers are in turn backed up via the network to tape via a separate backup server and LTO4 tape library.
Using the cluster
Members of the Faculty of Translational Medicine using our sequencing equipment have priority access to the High Performance Computing cluster.
Before you can use the cluster, you will need an account.
Once you have an account, you will be able to access your sequencing data securely 24/7 from anywhere in the world.
Contact
Don Riyage don.s.lokuadassuriyage@kcl.ac.uk
Don sets up accounts and supports researchers using the cluster.