Middleware Benchmarks

descriptionRAM (MB)CPU typedisk typepower consumption w/o disk (W)power consumption incl. disk (W)virtual or physical systemweb server softwaredb softwareoscommitcommentresult (s)
OPTIPLEX 790 SF w/ APC4096Corei3-2100 @3.1 GHzSATA???virtualapache2+phpmysqlLinux02c1ea557b927b5c7f90 0.13 w/ APC1024AMD 64 @1 GHzSCSI??physicalapache2+apcmysqlLinuxbaa7ab92bf87e0479c97 0.6 w/ APC1024AMD 64 @1 GHzSCSI??physicalapache2+apcmysqlLinuxfe69f572bff79a13a62a 0.61 w/ APC1024AMD 64 @1 GHzSCSI??physicalapache2+apcmysqlLinux0d60302e0fb251fe2918 0.58 w/o APC1024AMD 64 @1 GHzSCSI??physicalapache2+apcmysqlLinuxe4b1f69548f3e09ff837 0.811 w/o APC1024AMD 64 @1 GHzSCSI??physicalapache2+apcmysqlLinux2340233b5bb05da7d589 0.9 w/ APC512QEMU Virtual CPU version 0.12.3 @3.4 GHzn/a??virtualapache2+apcmysqlUbuntu 12.04.4 LTS380e084c0f8ad538dabd2014-01-12, new server, vserver now0.005
Seagate dockstar w/ usb stick128Marvel @1.2 GHzUSB2/flash?3.4physicalapache2+apcmysqlDebian Squeeze2340233b5bb05da7d589 2.6
Seagate dockstar w/ 3.5„ HD128Marvel @1.2 GHzUSB2/3,5“ HD3.1?physicalapache2mysqlDebian Squeezefe69f572bff79a13a62a 1.7
Seagate dockstar w/ 3.5„ HD128Marvel @1.2 GHzUSB2/3,5“ HD3.1?physicalapache2+apcmysqlDebian Squeeze2340233b5bb05da7d589 2.6
IOmega iConnect w/ 2.5„ HD256Marvel @1 GHzUSB2/3,5“ HD?5physicalapache2+apcmysqlDebian Squeeze0f6004bbec9c4fec027a 2.12
Seagate dockstar w/ 3.5„ HD w/o APC128Marvel @1.2 GHzUSB2/3,5“ HD3.1?physicalapache2mysqlDebian Squeeze2340233b5bb05da7d589 3.2
Neoware ThinClient EON 4300s512Via Ezra @800 MHzUSB1.1/disk19?physicalapache2mysqlUbuntu Lucid2340233b5bb05da7d589 1.43
Neoware ThinClient EON 4300s w/ APC512Via Ezra @800 MHzUSB1.1/disk19?physicalapache2+apcmysqlUbuntu Lucid2340233b5bb05da7d589 1.4
Foxconn NT525 w/ APC1024Intel Atom 525 DualCore @1.8GhzMicroSD Cheap&Slow?13physicalapache2+apcmysqlDebian Squeeze2340233b5bb05da7d589 0.71
Foxconn NT525 w/o APC 1024Intel Atom 525 DualCore @1.8GhzMicroSD Cheap&Slow?13physicalapache2mysqlDebian Squeeze2340233b5bb05da7d589 0.87
VMware w/o APC 128Intel Core 2 Duo @2.4Ghz (only 1 Core in VM)2.5„ SATA??virtualapache2mysqlDebian Squeeze8b65fa8ba25f149d7724 0.25
VMware w/ APC 128Intel Core 2 Duo @2.4Ghz (only 1 Core in VM)2.5“ SATA??virtualapache2+apcmysqlDebian Squeeze8b65fa8ba25f149d7724 0.18
Synology NAS DS-109+ 512Freescale PowerPC MPC8533 @1,06Ghz3.5„ SATA1221physicalapache2mysqlLinux 1.17
Thinkpad Edge XAMPP Win7 w/ APC 2048AMD II Neo K325 DualCore @1.3GHz2.5“ SATA?12physicalapache2+apcmysqlWindows7 x64 0.84
Thinkpad Edge XAMPP Win7 w/o APC2048AMD II Neo K325 DualCore @1.3GHz2.5„ SATA?12physicalapache2mysqlWindows7 x64 0.98
PC-Engines Alix3d2256AMD-Geode LX800-500MHzCF4GB +8GB-USB-Stick<5<5physicalapache2+apcmysqlDebian Squeeze 2.7
HP Proliant L36N1024AMD II Neo Dual-Core @1.3GHz3.5“ SATA Raid12530physicalapache2+phpmysqlFreeBSD 9 0.32
RaspberryPi256ARM11 @700 MHz2 GB SD-Card (Medion)n/a2.5physicalapache2+apcmysqlDebian Squeeze 3.46
RaspberryPi256ARM11 @700 MHz2 GB SD-Card (Medion)n/a2.5physicalapache2+apcmysql2012-07-15-wheezy-raspbian0d60302e0fb251fe2918 1.91
RaspberryPi128ARM11 @700 MHz2 GB SD-Card (Medion)n/a2.5physicalapache2+apcmysql2012-07-15-wheezy-raspbian0d60302e0fb251fe2918RAM setting via raspi-config memory split 1.91
Intel Atom1024Intel(R) Atom(TM) CPU 330 @ 1438.850MHz (underclocked)2.5„ SATAn/a23physicalapache2mysqlUbuntu Precise Pangolin 0.47
Banana Pi1024 Allwinner A20, ARM Cortex-A7, Dual-Core, 1GHz8GB SD-Card Sandisk Extremen/a2physicalapache2+apcmysqlBananian 14.09 (Debian 7) 0.59
Hummingboard i2eX1024i.MX6(ARM7) Dual-Core @1GHz16GB Samsung SD-Card, 64GB Kingston mSATA SSDn/a2.3 (3.7 peak)physicalapache2+apcmysqlIgor Pečovnik Debian Wheezy v2.28e0a5a1839bd3c1df005 0.35

And how are the results being retrieved?

  • import the demo dataset found in our git repo
  • run this query:
   http://<server>/<path to middleware>/data/a301d8d0-903b-1234-94bb-d943d061b6a8.json?from=1287541654974&to=1288488900567&tuples=1000&debug=5
  • find the reading somewhere a few lines above the bottom that looks something like this:
debug: {

     time: <benchmarkresult>

This is the time that passed while processing the request.


  • try to make sure there is as little extra load as possible on the system while the request is performed
  • give the request several runs; all (or most) requests should give more or less identical results; in case the results differ, submit an average
  • use the first 20 characters of the entry in „commit-hash“ for the commit field

Not happy with your results?

Give performance a look!

development/benchmark.txt · Zuletzt geändert: 2015/01/10 15:48 von hasselh