Hi Michael, On 2011-04-21, at 4:38 PM, Michael Hunger wrote:
> Bob, > > I don't know if you have already answered these questions. > > Which JDK (also version) are you using for that, what are the JVM memory > settings? Sun's java 1.6 patch level 24 (I think I'll have to confirm on Monday) Memory settings... from the top of my head... there's a GC setting that I can't recall, and the min and max heap is set to 12GB (which it never comes close to) > > Do you have a profiler handy that you could throw at your benchmark? (E.g. > yourkit has a 30 day trial, other profilers surely too). Monday for that too. > > Do you have the source code of your tests at hand? So we could run exactly > the same code on our own Linux systems for cross checking? If it's useful I can probably extract something, starting Monday :-) > > What Linux distribution is it, and 64 or 32 bit? Do you also have a disk > formatted with ext3 to cross check? (Perhaps just a loopback device). Ubuntu 10.10 64bit. The machine has been set up as ext4, I'll see what I can scrounge up for ext3. > > How much memory does the linux box have available? 16GB > > Thanks so much. > Thank you. Cheers, Bob > Michael > > Am 21.04.2011 um 21:53 schrieb Bob Hutchison: > >> >> On 2011-04-20, at 7:30 AM, Tobias Ivarsson wrote: >> >>> Sorry I got a bit distracted when writing this. I should have added that I >>> then want you to send the results of running that benchmark to me so that I >>> can further analyze what the cause of these slow writes might be. >>> >>> Thank you, >>> Tobias >> >> That's what I figured you meant. Sorry for the delay, here they are: >> >> On a HP z400, quad Xeon W3550 @ 3.07GHz >> ext4 filesystem >> ----- >> >>>> dd if=/dev/urandom of=store bs=1M count=1000 >> 1000+0 records in >> 1000+0 records out >> 1048576000 bytes (1.0 GB) copied, 111.175 s, 9.4 MB/s >>>> dd if=store of=/dev/null bs=100M >> 10+0 records in >> 10+0 records out >> 1048576000 bytes (1.0 GB) copied, 0.281153 s, 3.7 GB/s >>>> dd if=store of=/dev/null bs=100M >> 10+0 records in >> 10+0 records out >> 1048576000 bytes (1.0 GB) copied, 0.244339 s, 4.3 GB/s >>>> dd if=store of=/dev/null bs=100M >> 10+0 records in >> 10+0 records out >> 1048576000 bytes (1.0 GB) copied, 0.242583 s, 4.3 GB/s >> >> >>>> ./run ../store logfile 33 100 500 100 >> tx_count[100] records[31397] fdatasyncs[100] read[0.9881029 MB] >> wrote[1.9762058 MB] >> Time was: 5.012 >> 19.952114 tx/s, 6264.365 records/s, 19.952114 fdatasyncs/s, 201.87897 kB/s >> on reads, 403.75793 kB/s on writes >> >>>> ./run ../store logfile 33 1000 5000 10 >> tx_count[10] records[30997] fdatasyncs[10] read[0.9755144 MB] >> wrote[1.9510288 MB] >> Time was: 0.604 >> 16.556292 tx/s, 51319.54 records/s, 16.556292 fdatasyncs/s, 1653.8523 kB/s >> on reads, 3307.7046 kB/s on writes >> >>>> ./run ../store logfile 33 1000 5000 100 >> tx_count[100] records[298245] fdatasyncs[100] read[9.386144 MB] >> wrote[18.772287 MB] >> Time was: 199.116 >> 0.5022198 tx/s, 1497.8455 records/s, 0.5022198 fdatasyncs/s, 48.270412 kB/s >> on reads, 96.540825 kB/s on writes >> >> procs -----------memory---------- ---swap-- -----io---- -system-- ----cpu---- >> r b swpd free buff cache si so bi bo in cs us sy id wa >> 1 2 0 8541712 336716 3670940 0 0 1 7 12 20 4 1 95 >> 0 >> 0 2 0 8525712 336716 3670948 0 0 0 979 1653 3186 4 1 60 >> 35 >> 1 2 0 8525220 336716 3671204 0 0 0 1244 1671 3150 4 1 71 >> 24 >> 0 2 0 8524724 336716 3671332 0 0 0 709 1517 3302 4 1 65 >> 30 >> 0 2 0 8524476 336716 3671460 0 0 0 1033 1680 69342 5 7 59 >> 29 >> 0 2 0 8539168 336716 3671588 0 0 0 1375 1599 3272 3 1 70 >> 25 >> 1 2 0 8538860 336716 3671716 0 0 0 1157 1594 3097 3 1 72 >> 24 >> 0 1 0 8541340 336716 3671844 0 0 0 1151 1512 3182 3 2 70 >> 25 >> 0 1 0 8524812 336716 3671972 0 0 0 1597 1641 3391 4 2 72 >> 22 >> >> >> _______________________________________________ >> Neo4j mailing list >> User@lists.neo4j.org >> https://lists.neo4j.org/mailman/listinfo/user > > _______________________________________________ > Neo4j mailing list > User@lists.neo4j.org > https://lists.neo4j.org/mailman/listinfo/user ---- Bob Hutchison Recursive Design Inc. http://www.recursive.ca/ weblog: http://xampl.com/so _______________________________________________ Neo4j mailing list User@lists.neo4j.org https://lists.neo4j.org/mailman/listinfo/user
