4 Replies Latest reply on Jan 24, 2013 3:33 PM by CarlSchwarz

    FMS oddly much slower on better hardware - Mac Minis- test results

    mikem

      Hi Developer Forum,

       

      I am hoping someone can shed some light on a most puzzling set of results. I am encouraging my client to move from hosting their database on a 2007 Mac Mini, to a 2011 Mac Mini. The new machine's specs are of course better in all respects, yet the database runs much slower. Some quick facts:

       

      Old setup

      mid-2007 Mini

      1.83GHz Core 2Duo, 2GB RAM

      attached to internet via 30Mbit up/15Mbit down Comcast connection (27 ms ping)

      OS X 10.6 & FMS v10

       

      New setup

      2011 Mini

      2.3GHz i5, 4GB RAM

      attached to intenet via 90Mbit up/down connection (75 ms ping)

      OS X10.6 & FMS v10 trial (1st round of tests)

      OS X 10.7 & FMS v11 trial (2nd round)

       

      Database: 28 files

       

      All testing takes place by remote connection via a 2.5Mbit DSL line, on a 2009 iMac with FMP v11.0.4

       

      Sample results

      Open database to main menu:

      Old: 0:52

      New: 1:34

       

      Quit database:

      Old: 0:18

      New: 0:25

       

      Results on new server were about the same whether FMS v10 or v11 was used.

      Results for Finds, etc have a similar pattern (I can come up with some numbers if that is helpful).

      To bring up the new database, I just took the 28 .fm7 files, uploaded using the FMS interface, and begain testing. No tweaking or other prep. I have run the tests multiple times without shutting down the server, so any caching etc should have occurred, I would think. Web Publishing and ODBC is off on all servers.

       

      Can anyone shed some light on these results? I feel pretty goofy telling the client, "To optimize performance, just stick with your 5 year old server with half the RAM!" Could the increased latency of the new setup (located on the other side of the country) be the issue?

       

      Mike May

      p.s. for reference, my results when files are accessed off local hard drive:

      Open database:

      0:18

      Quit database:

      0:03

       

      p.p.s Apologies if this should have gone into the Performance Tuning forum. It seems less "tuning" and more basic operation.