cryoEM DAQ Horizontal

cryoEM DAQ

 

 

Simplified

 

  • Actors
    • cryoEM Operator is the person(s) operating the TEM at an authorised period for data acquisition
    • Local Administrator Account to control login authorisation based on who should have access
    • Data Acquisition Collaborator(s) are people who have SLAC accounts whom the cryoEM Operator may want to help operate the TEM
    • Researchers are users who would use the compute and storage facilities at SLAC to process the data that has been acquired.
  • Resources
    • Local Windows Machine
      • Connected to SLAC Active Directory
      • Runs MinSec (virus scanner, cyber security backdoor, centralised logging and authentication)
      • Has RAID Storage?
    • CryoEM Windows Servers
      • Standard FEI and Gatan servers used for data collection and TEM control
      • Data stored locally per TEM
    • CryoEM Data Acquisition Servers
      • Unix Docker Swarm / Kubernetes Nodes that handles data movement and processing pipelines
      • Utilises Apache Airflow to orchestrate workflows
      • Launches jobs on GPU and CPU farm clusters that contain required software (motioncor2, ctffind, dogpicker etc)
      • Shares storage with CryoEM Windows Servers via a CIFS mount for each and every TEM.
    • GPFS LTS Storage + TSM HSM Tape
      • Long term data storage (multi petabyte) using clustered parallel file system
      • Filesystem mounted on all (Unix) Nodes
      • Automatic policies to stage unused/not-hot data to tape for cheaper storage needs
    • Bullet CPU Cluster
      • CPU based cluster (5,000 cores); infiniband + 10GbE
    • GPU Cluster
      • GPU based cluster (100 gpus)
    • Interactive Login Hosts
      • 'Head' nodes where Researchers log in to access compute resources
    • Data Transfer Nodes
      • High (network) speed nodes where Researchers log into access the data
  • Researchers
    • Will collect data from the CryoEM Windows Servers (using SerialEM/EPU etc) either
      • Onsite: in building 6
      • Remotely: connect via FastX (only available when their experiment is active)
    • Once data collection sessions is in process/complete, researchers can
      • submit jobs to the GPU and CPU clusters via the Interactive Login Hosts
      • copy data via the Data Transfer Nodes

 

Further Information

 

 

 

 

 

  • No labels