> 1. Did I get this right: BioMart uses a "Data-Warehouse"-architecture > which means that the user doesn't access the "original databases" but > modified copies of them on the biomart-servers?
Yes, you're right. > What does the term > "federated databases" mean in this context (which is used in many of the > presentations on the biomart website)? It means that although it works off modified copies (which it calls marts), you don't have to have all those marts stored on the same server. Instead, many different people can give you access to the marts on their own servers, and the query system will bring together the results it needs from each one at query execution time. > 2. How do you ensure that BioMart does not shut down? Do you use a "load > balancer"-server to sort crashed servers out? Or do you have some kind > of backup-servers? Or can BioMart access the original databases, if the > "copies on the biomart servers" don't work? BioMart has no load-balancing or fail-over or fall-back capacity - if it crashes, you just have to restart it. It also cannot read raw data - it has to work off the modified marts, so if the database server containing those marts is down, BioMart can't access that data until the database server comes back up or it is reconfigured to use an alternative source instead (reconfiguration requires a restart). Hopefully this helps! thanks, Richard > Thanks very much, > Stephen > > -- Richard Holland, BSc MBCS Finance Director, Eagle Genomics Ltd T: +44 (0)1223 654481 ext 3 | E: [email protected] http://www.eaglegenomics.com/
