Free-DC
04-28-2010, 02:39 AM
...due to improvements we are currently working on to process the tiny WUs on the server in the future. The background is that batches of very short WUs cause severe server stress regarding the number of connections and MySQL requests despite our good server hardware specifications. Since these small WUs will re-occurr in future projects as well, we need to do something about this. At present, we have implemented a server response delay which is causing many machines to sit idle exclusively once these small WU batches enter the system - a situation which clearly is unacceptable. These WUs are usually consisting of plasmids or phages with very short DNA sequences and are consequently processed within just a few minutes - sometimes even shorter depending on your hardware - and therefore the server contact frequencies are massively increased. We could solve the issue by just adding a couple of servers as put into practice by some of the bigger DC projects. Unfortunately, our current project budget does not allow for such a solution, so we decided to process these small WUs on one or two of the server cores in the future. Michael.
More... (http://www.rnaworld.de/rnaworld/forum_thread.php?id=49)
More... (http://www.rnaworld.de/rnaworld/forum_thread.php?id=49)