About: Informatics

General computing infrastructure

The general computing infrastructure is diagrammed in Fig. 1. The major classes of computational facilities include:

  • Server room with dedicated cooling, local emergency power backup (UPS), and connection to building backup generator (350sf)
  • Central servers and storage housed in the above
  • Informatics personnel office and resource library space (900sf)
  • Informatics workstations (8)
  • Laboratory instrument workstations (15)
  • General desktop workstations (50)

Summary of computational resources schematized in Fig. 1

Our computational facility is comprised of ~346 CPU cores and ~223TB of online data storage capacity. We host a primary computational cluster consisting of 116 CPU slots with a combination of Intel Xeon E5410 2.33GHz and AMD dual core Opteron 880 processors. An NFS (Network File System) server with 23 TB capacity acts as the Sun Grid Engine cluster head node. Inactive data is staged on a Sun Fire X4500 server with 34TB of SATA storage.

A secondary computational cluster is reserved for ABI SOLiD next-generation sequencing offline analysis. This cluster is composed of twelve systems each with 8 Intel Xeon X5550 2.67GHz processor cores and 48GB RAM. Attached to this cluster is 34TB of 7.2K rpm SAS storage.

The HiSeq Illumina instrument has two pipeline servers for temporary storage and secondary analysis. The pipeline servers each have 5TB of internal SATA storage, eight X5355 2.66GHz processor cores, and 16GB RAM. The HiSeq Illumina storage server has 48TB of 7.2K RPM Near-Line SAS disks, eight 2.66GHz cores and 48GB RAM.

Figure 1.

General Schematics of Computational architecture showing various compute clusters dedicated to specific tasks. The infrastructure is supported by 2 servers, a MySQL server and a Laboratory Information Management System (LIMS) server for samp[le and data tracking and generating reports on demand.

Additional files are stored on NFS servers. A virtual host server supports various virtualized application servers. Other web servers include an external server for public data release, as well as web servers used for visualization, bioinformatics database, and LIMS applications. All networked file systems, databases, and home directories are backed up using Tivoli Storage Manager software to a robotic Dell ML6000 library with four LTO4 tape drives. The library is capable of storing 132 LTO4 tapes, with a total possible internal storage capacity of 211.2TB. With an inventory of 500 active tapes, this gives us a total of 800TB offline storage.