1106 A petabyte-scale data platform for modern clinical research at UTSW Medical Center.pdf

1、A Petabyte-scale Data Platform at UTSW Medical CenterCong Wang&Liqiang Wang UTSW1Who We Are2Medical Research at UTSWTo be the nations premier institution for clinical research committed to novel,impactful science,and the translation of discoveries into practice to improve the lives of patients and c

2、ommunities.Has been fostered a rich collaboration among faculties,biostatisticians,bioinformaticians,AI computer scientists,and the domain experts that supports clinical research projects.BioHPC Services Empowers UTSW research by giving our researchers easy access to the HPC infrastructure needed to

3、 tackle their most pressing life science questions.Covers all aspects of HPC for users,from technical jobs such as cluster expansion and storage management to carrying out collaborative work on computational aspects of research projects.Storage32013 240TBToday 150PB+Strategies4Cost Performance Capac

4、ity SecuritySupportScience Procedures5BioHPC-Demo6HPC7Upload dataInstall APPscodingSubmit jobsDownload dataDown steamChangesreproduceWhat is Astrocyte?A platform for workflows/pipelinesFeatures:Standardized workflows (Developed by experts)Good reproducibility (Restrict version control policy)Easy-to

5、-use web interface (No cluster knowledge required)Online documents&results visualization*Layout of BioHPC9Off-campusCampusQTSCPU nodesGPU nodesLogin node(Nucleus005)CUH1.6Tb/s3.5PB GL6/workTwo GPFS clusters:keep production backup14PB ESS5000/project/archive/home28.9 PB GL6S 90GB/s/work/archiveHuawei

6、 NVME/home220PB+1PB Flash(DDN)new DDN 400GB/s/projectvnc.bhpc.sweduUsers WifiUsers(VPN)ESXI Hosts(2 HPE,?Dell)ESXI Hosts(8 HPE,?Dell)ALL BioHPC VMsSuper node(Nucleus001)Nucleus001Nucleus005 10 x lower latencyUsers WiredIR VPN Server(16 x 100Gb/s)Users(External)lamellabackup001(IB)200Gb/s x40 x10IR C