Ivan, >> Can you describe that experiment in more details?
Nature of a company: 70+ XML files ranging from 1 MB to 500 MB. Small companies have Sedna dbs of 200+ MB, while large companies have 2+ GB. Performance measurement: Using log file data from the front end services. I don't have an values to share, sorry. The following describes the overall Amazon EC2 XML experience. The original concept was to dynamically build unique Amazon EC2 small instance for each company. Start & stop services based on usage. The issue with EC2 small instances is low/limited bandwidth. The second approach was a single instance on a large box, with a unique catalog for each company. The test was with eXist-db, and as the db grew large performance problems became obvious. (The last time I checked, there was 60 catalogs with a db size over 30 GB.) The third approach was to dynamically create dbs. Sedna seemed much better suited than eXist-db for this activity. I'm sorry I don't have any hard empirical data at the moment. I've been switching between eXist-db and Sedna. In about 30-60 days, I should have enough data and time to try several experiments (500 GB versus 120 unique dbs). We have about 60 unique companies signing up each month. Thank you very much for the suggestion on the OS and Sedna tuning. Malcolm -----Original Message----- From: Ivan Shcheklein [mailto:[email protected]] Sent: Wednesday, December 01, 2010 12:22 PM To: Malcolm Davis Cc: sedna-discussion Subject: Re: [Sedna-discussion] SEDNA Message: ERROR SE4211: maximum number of databases Malcolm, e. We tried a single database instance with a separate catalog for each company, but performance started to suffer. Can you describe that experiment in more details? * Did you try to increase buffers? * What is the total size of the data in Sedna? * How did you measure performance - I mean what kind of queries/updates did you use? * What was the database structure - one collection/multiple collections/multiple stand-alone documents? Does the concept seem reasonable? Probably. No one tried it so far, AFAIK. From the other side, if there are few clients at every moment why the single-database solution doesn't work? Are there any limitations or issues with using Sedna in this way? The only one strong limitation is that "Sedna's database" is very heavy - buffers, a lot of IPC objects (e.g. semaphores), etc. You will need to rebuild Sedna and most likely to tune your OS to run so many databases at once. Ivan ------------------------------------------------------------------------------ Increase Visibility of Your 3D Game App & Earn a Chance To Win $500! Tap into the largest installed PC base & get more eyes on your game by optimizing for Intel(R) Graphics Technology. Get started today with the Intel(R) Software Partner Program. Five $500 cash prizes are up for grabs. http://p.sf.net/sfu/intelisp-dev2dev _______________________________________________ Sedna-discussion mailing list [email protected] https://lists.sourceforge.net/lists/listinfo/sedna-discussion
