8 Replies Latest reply on Mar 26, 2014 7:32 AM by Vincent_L

    Mac Pro 2014 vs Mac Mini 2013 with Pegasus RAID

    taylorsharpe

      I just got a new Mac Pro in - yahoooo. Its a 6 core machine with 32 Gigs of RAM and 1 TB storage. The current system is running a Mac Mini from 2013 attached to a Promise RAID (Pegasus R4) and the live databases are on the RAID. Both systems are Mac OS X Mavericks server running FileMaker Server 13. The initial test of a large SQL sum returned an answer on the Mac Pro returned an answer in 2950 miliseconds and on the Mac Mini it was 4500 miliseconds. Running the test several times on each machines improved the results a lot and both machines got to where they were returning the same answer in about 3000 miliseconds. So for my first test, a SQL sum, the Mac Pro for an individual process is at best about 25% faster. I'll do some more testing, but always interesting to compare.

        • 1. Re: Mac Pro 2014 vs Mac Mini 2013 with Pegasus RAID
          LSNOVER

          I would not expect much difference with a single user.  Typically the beefier hardware with more CPU cores hold up better to more users.  AFAIK, Filemaker is still executing code for a specific user in a single thread on a single CPU.  You've addressed the biggest bottleneck with the fast drive on the Mini.  Certainly interested in other comparisons.  Jealous of the Mac Pro.  ;-)

           

          Keep us posted!

          • 2. Re: Mac Pro 2014 vs Mac Mini 2013 with Pegasus RAID
            taylorsharpe

            You are correct, Lee, in that FileMaker only is executing a single thread on a single core for a given user as far as I can tell.  But I just had this new machine and wanted to compare a simple task, SQL sum, to see how the E5 processor compares to the i7 processor.  Since it is a brand new machine and I didnt' have much time to test other things and I expect the Mac Pro will shine more when loaded down with a lot of users.  I would love to have seen the E5 processor kick the i7's butt, but that probably is not the case on any single thread and user for FileMaker.  At first glance, for small offices, a Mac Mini server probalby is a much better match to the SOHO needs if you have SSDs or  fast RAID. 

             

            Lee... if you can think of some good suggestions to run to make some comparisons, I'm thinking about it.  Such as importing a large file with lots of indexed fields and running container field import and exports and large calculation sums over and over in a script and run them for hours to compare which does better.  Maybe run several scripts at the same time and try to give it a test load of users. 

             

            Well, its not my Mac Pro, but it is fun giving it a whirl and testing things out!  I'll let you know more as I test it. 

            • 3. Re: Mac Pro 2014 vs Mac Mini 2013 with Pegasus RAID

              Hi Taylor,

               

              I'd be interested to see what performance you can get when the DB is residing on the new MacPro's internal SSD's instead of being on the Pegasus RAID.

               

              For as long as i can remember, we've been told that read and write speed to the stored DB files are one of the most important limitations on performance.

               

              Can you possibly run a test to measure that component of performance?

               

              Cheers,

               

              John

              • 4. Re: Mac Pro 2014 vs Mac Mini 2013 with Pegasus RAID
                taylorsharpe

                The Mac Mini I have does not have SSDs, only slow 2.5" 5400 rpm drives.  But, the Pegasus RAID has faster MB/s than an SSD (maybe not as fast of latancy though).  That is why I tested speeds on the RAID because if I tested on the 5400 rpm drive, performance would be dismal. 

                 

                You are correct that read and write speeds on a database are very important and sometimes more important than the cpu.  The weak link on a Mac Mini has most often been the drives, but with SSDs that has improved a lot.  But SSDs stlil are not up to the MB/s throughput you can get with a Promise RAID.  And the RAID has redundancy which is typically important to a production server. 

                 

                If I get a hold of a Mac Mini with SSDs, I'll run a comparison.  This is going to be interesting making such comparisons because I know we all want to make good recommendations to our clients. 

                • 5. Re: Mac Pro 2014 vs Mac Mini 2013 with Pegasus RAID
                  Vincent_L

                  Taylor,

                   

                  Forget about the mac mini, I think John asked Mac Pro 2014 internal SSD vs Mac Pro 2014 database hosted on Promise.

                  My guess : either the same or Mac Pro's internal SSD faster

                  • 6. Re: Mac Pro 2014 vs Mac Mini 2013 with Pegasus RAID
                    GordonShewach

                    I'm interested in this, too, Vincent. I have a Mac Pro on order (April delivery) and planning on benchmarking:

                     

                    • Internal SSD

                    • Promise Pegasus2

                    • 2 Mercury Accelsior PCIe SSDs RAIDed in a Helios 2 PCIe Thunderbolt Expansion Chassis

                     

                    My client already owns the Accelsior cards and is willing to buy a Promise Pegasus2 ($1,500 with 14-day return policy) and a Helios 2 ($480 with 30-day return policy but 15% restocking fee). We'll return the loser in the benchmark tests.

                    • 7. Re: Mac Pro 2014 vs Mac Mini 2013 with Pegasus RAID
                      taylorsharpe

                      I have a Mac Pro 2008 with 8 Gigs of RAM and a Caldigit RAID card (RAID 5) with 4 drives.  I could test that, but it is a configuration probably few people have. 

                       

                      I heard good things about the Accelsior cards too.  I'm very interested in its comparison with Pegasus Thunderbolt 2!

                      • 8. Re: Mac Pro 2014 vs Mac Mini 2013 with Pegasus RAID
                        Vincent_L

                        For me,

                        All wil be the same, Filemaker is CPU bound (of course if you don't use HD) for most tasks. I Hosted my database on ram disk = fastest storage possible, and did not experience noticeable performance gains over SSD.

                         

                        Moreover now 13 has big Ram cache, so I'd max ram fisrt (according to solution * 2)