> What is Destiny?
Introduction
Blade servers are a relatively new technology that has captured industry focus because of its modular design, which can reduce cost with a more efficient use of valuable floor space, and its simplified management, which can help to speed up such tasks as deploying, updating, and troubleshooting hundreds of blade servers. All this can be done remotely with one graphical console. In addition, blade servers provide improved performance by doubling current rack density.
Specifications
- The Hartwell Center IBM Linux cluster has 1 management node, 140 compute nodes and 1 storage node:
- Each Hartwell Center BladeCenter chassis (10) holds 14 blades.
- Each of the 140 blades has 2, 2.4 GHz/400Mhz Xeon Processors for a total of 280 processors.
- Each of the 140 blades has 2.5 GB of memory.
- Each of the 140 blades has a 40GB hard drive.
- The Management, Compute and Storage node use Red Hat Advanced Server as the operating system.
- The function of the Storage node is to interact with the Hartwell Center storage area network, (SAN). The SAN has 9 TB of storage.
- We use the Linpack benchmark, a performance measure for ranking the computer systems in the TOP500 Supercomputer sites. http://www.top500.org/list/2003/11
- As a measure of computer speed, a gigaflop is a billion floating-point operations per second).
- In computers, FLOPS are floating-point operations per second. Floating-point is, according to IBM, "a method of encoding real numbers within the limits of finite precision available on computers." Using floating-point encoding, extremely long numbers can be handled relatively easily. A floating-point number is expressed as a basic number or mantissa , an exponent, and a number base or radix (which is often assumed). The number base is usually ten but may also be 2. Floating-point operations require computers with floating-point registers . The computation of floating-point numbers is often required in scientific or real-time processing applications and FLOPS is a common measure for any computer that runs these applications.
- In larger computers and parallel processing, computer operations can be measured in megaflop s, gigaflop s, and teraflop s. Some computer scientists have at least begun to think about petaflop s.
Tools
What tools will be available on the IBM Linux Cluster?
- Current Applications: Cluster Amber, Clustal-W, mpiBlast, MeMe, and Hmmer, CNS, Consensus.
- Compilers: Portland Group Cluster Development Kit: parallel FORTRAN, C and C++
- Job Scheduling and Queuing: LSF is a distributed load sharing and batch queuing software that manages, monitors and analyzes the workload on a network of heterogeneous computers. Each computer on the network is treated as a node and LSF is able to choose which node is best suited to run your job by measuring Load averages and other system resources.
Why is this important to us?
Using a 16-cpu onyx2 it took 11 days to run a 15,000 sequence BLAST search. On April 16, 2003 , the same search took only 4 days, 14 hours, 32 minutes and 52 seconds on a small test cluster. And on August 5, 2003 the search was run on the IBM Linux cluster. It took 2 hours and 57 minutes.
The cluster allows us to do science otherwise impossible; it allows us to perform cpu-intensive calculations in the fields of Structural Biology, Bioinformatics, Statistics, Image Analysis, etc. Access to computational horsepower of this magnitude is increasingly important as we are inundated with volumes of large datasets from gene expression, aCGH, SNP, genome, and proteomics studies.
