Mascot has been designed to be embarrassingly parallel, which means that each search can be divided to run in parallel on multiple processors. Support for parallel execution is built-in to Mascot, and does not require any special operating system or grid engine. If you wish, you can license the software and set up a cluster using your own hardware. On this page, we describe the other option: a turn-key system with Mascot pre-installed on reliable, high performance IBM hardware.
Hardware
Any current IBM server that uses Intel or AMD processors can be used as the hardware platform. For a small cluster, up to around 6 processors, the most cost-effective solution is to use a number of dual processor servers, either tower or rack mount. For larger systems, we recommend the IBM BladeCenterTM, which is more compact and easier to manage than a large number of standard servers.
Mascot licensing is socket based. That is, if a processor has multiple cores, Mascot will use all of them for searching. For a given architecture, search speed (or throughput) goes as the processor clock speed multipled by the total number of cores.
Only the processors used for searching require a Mascot licence. It is often a good idea not to run searches on the master (head) node of the cluster, but leave it free to run the web server, handle database updates, and generate reports. This makes the server very responsive even when there are several searches running, using all the processor time on the search nodes. Having a non-searching master node also gives you a spare node, in case one of the search nodes has a hardware failure.
Result files are stored on the master node, so this needs access to plenty of disk storage. Search nodes only need local storage for program files and the compressed sequence database files; a single hard drive of modest size is usually sufficient. All nodes should have at least 2 GB RAM.
Software Link: Mascot Cluster - Spectroscopy Software
