Oninit cdi / Archive
Talk to an expert →

Buffer Wait Ratio question

Posted in 2000

Topics: General Discussion

DWillems — — source: Informix-list mailing list archive (1991-1998)

Hello, Below a list how our BR changes. (The first was before the changes!) runtime bufwaits pagreads bufwrits Buffer Wait Ratio 27 days 12226304 26192222 34465407 20,16% 1 hour 38500 575811 139906 5,38% 43567 613795 153733 5,68% 53252 740991 221636 5,53% 71902 769515 235190 7,16% 79256 978657 285396 6,27% 2 hours 103574 1351251 360387 6,05% 133874 1886781 602366 5,38% 19 hours 676756 35230173 3575865 1,74% 679187 35272242 3591536 1,75% 794928 36819299 3791788 1,96% 23 hours 1002939 37081926 4754493 2,40% 1978449 45031222 8600347 3,69% 46 hours 2362859 41813844 9211124 4,63% 2684742 41699880 9718835 5,22% 67 hours 3230158 35047802 11945757 6,87% 3653543 36666572 13235418 7,32% 90 hours 4076131 33863611 16229932 8,14% Questions about this: a) since our 90 BaaN never do start exceptional jobs and the nightjob always does the same routines I expected to see a stable BR after a few days. Can someone try to explain this, please ? b) I also thought that the 3 BR parameters are running totals. But pagreads sometimes lowers. Maybe this is the answers to my first question, but is this a normal behaviour? Why is this the only that behaves in this way? Thanks -- *** remove spam twice when replying ***

↪ replying to DWillems
Art S. Kagel — — source: Informix-list mailing list archive (1991-1998)

DWillems wrote: > > Hello, > > Below a list how our BR changes. (The first was before the changes!) > > runtime bufwaits pagreads bufwrits Buffer Wait Ratio > 27 days 12226304 26192222 34465407 20,16% > 1 hour 38500 575811 139906 5,38% > 43567 613795 153733 5,68% > 53252 740991 221636 5,53% > 71902 769515 235190 7,16% > 79256 978657 285396 6,27% > 2 hours 103574 1351251 360387 6,05% > 133874 1886781 602366 5,38% > 19 hours 676756 35230173 3575865 1,74% > 679187 35272242 3591536 1,75% > 794928 36819299 3791788 1,96% > 23 hours 1002939 37081926 4754493 2,40% > 1978449 45031222 8600347 3,69% > 46 hours 2362859 41813844 9211124 4,63% > 2684742 41699880 9718835 5,22% > 67 hours 3230158 35047802 11945757 6,87% > 3653543 36666572 13235418 7,32% > 90 hours 4076131 33863611 16229932 8,14% > > Questions about this: > a) since our 90 BaaN never do start exceptional jobs and the nightjob always > does the same routines I expected to see a stable BR after a few days. Can > someone try to explain this, please ? Yes it should stabilize. > b) I also thought that the 3 BR parameters are running totals. But pagreads > sometimes lowers. Maybe this is the answers to my first question, but is > this a normal behaviour? Why is this the only that behaves in this way? I see it but I don't believe it. Is it possible that the pagreads is wrapping around INT_MAX and back to zero? Considering the number of complaints I see in CDI about Baan queries doing sequential scans, I think that more likely. Try taking a snapshot hourly and see what happens and when. > Thanks > > -- > *** remove spam twice when replying *** Almost noone will despammify you email address here. Art S. Kagel

↪ replying to Art S. Kagel
DWillems — — source: Informix-list mailing list archive (1991-1998)

I'm monitoring pagreads and it drops without any visible reason, nobody runs any exceptional sessions. runtime dskreads bufreads bufwaits pagreads bufwrits BR 09 20:49 91.953.736 8.739.668 30.572.866 33.149.339 13,72% 09 21:14 92.244.184 8.768.787 * 28.792.723 * 33.684.247 14,04% 09 22:04 92.517.382 8.805.654 29.115.461 33.812.356 13,99% 09 23:11 93.047.195 8.862.940 29.694.772 34.301.424 13,85% 09 23:55 93.226.723 2.168.559.538 8.878.107 29.883.085 34.734.642 13,74% 10 00:23 93.549.084 2.173.973.691 8.907.593 30.237.529 35.211.662 13,61% 10 00:53 93.630.378 2.176.104.465 8.917.257 * 28.240.833 * 35.253.688 14,04% 10 01:47 93.769.743 2.180.545.039 8.935.174 28.394.699 35.324.085 14,02% 10 02:25 94.092.572 2.186.988.846 8.967.289 28.745.327 35.802.014 13,89% 10 02:39 94.409.402 2.189.645.800 8.986.628 * 27.921.840 * 36.301.578 13,99% 10 17:55 99.172.016 2.309.841.490 9.418.553 34.773.651 38.012.491 12,94% 10 19:21 99.566.236 2.324.843.948 9.475.500 35.204.615 38.551.209 12,85% 10 19:57 100.065.635 2.332.880.934 9.559.153 35.728.328 38.680.912 12,85% 10 22:08 101.334.085 2.358.091.318 9.738.567 37.073.884 38.918.770 12,82% Is it a good idea to replace pagreads by dskreads to calculate the BR ? Daniel Willems Art S. Kagel <kagel@bloomberg.net> wrote in message news:3A40F152.EC57303E@bloomberg.net... > DWillems wrote: > > > > Hello, > > > > Below a list how our BR changes. (The first was before the changes!) > > > > runtime bufwaits pagreads bufwrits Buffer Wait Ratio > > 27 days 12226304 26192222 34465407 20,16% > > 1 hour 38500 575811 139906 5,38% > > 43567 613795 153733 5,68% > > 53252 740991 221636 5,53% > > 71902 769515 235190 7,16% > > 79256 978657 285396 6,27% > > 2 hours 103574 1351251 360387 6,05% > > 133874 1886781 602366 5,38% > > 19 hours 676756 35230173 3575865 1,74% > > 679187 35272242 3591536 1,75% > > 794928 36819299 3791788 1,96% > > 23 hours 1002939 37081926 4754493 2,40% > > 1978449 45031222 8600347 3,69% > > 46 hours 2362859 41813844 9211124 4,63% > > 2684742 41699880 9718835 5,22% > > 67 hours 3230158 35047802 11945757 6,87% > > 3653543 36666572 13235418 7,32% > > 90 hours 4076131 33863611 16229932 8,14% > > > > Questions about this: > > a) since our 90 BaaN never do start exceptional jobs and the nightjob always > > does the same routines I expected to see a stable BR after a few days. Can > > someone try to explain this, please ? > > Yes it should stabilize. > > > b) I also thought that the 3 BR parameters are running totals. But pagreads > > sometimes lowers. Maybe this is the answers to my first question, but is > > this a normal behaviour? Why is this the only that behaves in this way? > > I see it but I don't believe it. Is it possible that the pagreads is > wrapping around INT_MAX and back to zero? Considering the number of complaints > I see in CDI about Baan queries doing sequential scans, I think > that more likely. Try taking a snapshot hourly and see what happens and > when. > > > Thanks > > > > -- > > *** remove spam twice when replying *** > > Almost noone will despammify you email address here. > > Art S. Kagel

↪ replying to DWillems
Art S. Kagel — — source: Informix-list mailing list archive (1991-1998)

Very strange. I've never seen it. Looks like a call to tech support is in order. Art S. Kagel DWillems wrote: > > I'm monitoring pagreads and it drops without any visible reason, nobody runs > any exceptional sessions. > > runtime dskreads bufreads bufwaits pagreads > bufwrits BR > 09 20:49 91.953.736 8.739.668 30.572.866 > 33.149.339 13,72% > 09 21:14 92.244.184 8.768.787 * 28.792.723 * > 33.684.247 14,04% > 09 22:04 92.517.382 8.805.654 29.115.461 > 33.812.356 13,99% > 09 23:11 93.047.195 8.862.940 29.694.772 > 34.301.424 13,85% > 09 23:55 93.226.723 2.168.559.538 8.878.107 29.883.085 > 34.734.642 13,74% > 10 00:23 93.549.084 2.173.973.691 8.907.593 30.237.529 > 35.211.662 13,61% > 10 00:53 93.630.378 2.176.104.465 8.917.257 * 28.240.833 * > 35.253.688 14,04% > 10 01:47 93.769.743 2.180.545.039 8.935.174 28.394.699 > 35.324.085 14,02% > 10 02:25 94.092.572 2.186.988.846 8.967.289 28.745.327 > 35.802.014 13,89% > 10 02:39 94.409.402 2.189.645.800 8.986.628 * 27.921.840 * > 36.301.578 13,99% > 10 17:55 99.172.016 2.309.841.490 9.418.553 34.773.651 > 38.012.491 12,94% > 10 19:21 99.566.236 2.324.843.948 9.475.500 35.204.615 > 38.551.209 12,85% > 10 19:57 100.065.635 2.332.880.934 9.559.153 35.728.328 > 38.680.912 12,85% > 10 22:08 101.334.085 2.358.091.318 9.738.567 37.073.884 > 38.918.770 12,82% > > Is it a good idea to replace pagreads by dskreads to calculate the BR ? > > Daniel Willems > > Art S. Kagel <kagel@bloomberg.net> wrote in message > news:3A40F152.EC57303E@bloomberg.net... > > DWillems wrote: > > > > > > Hello, > > > > > > Below a list how our BR changes. (The first was before the changes!) > > > > > > runtime bufwaits pagreads bufwrits Buffer Wait Ratio > > > 27 days 12226304 26192222 34465407 20,16% > > > 1 hour 38500 575811 139906 5,38% > > > 43567 613795 153733 5,68% > > > 53252 740991 221636 5,53% > > > 71902 769515 235190 7,16% > > > 79256 978657 285396 6,27% > > > 2 hours 103574 1351251 360387 6,05% > > > 133874 1886781 602366 5,38% > > > 19 hours 676756 35230173 3575865 1,74% > > > 679187 35272242 3591536 1,75% > > > 794928 36819299 3791788 1,96% > > > 23 hours 1002939 37081926 4754493 2,40% > > > 1978449 45031222 8600347 3,69% > > > 46 hours 2362859 41813844 9211124 4,63% > > > 2684742 41699880 9718835 5,22% > > > 67 hours 3230158 35047802 11945757 6,87% > > > 3653543 36666572 13235418 7,32% > > > 90 hours 4076131 33863611 16229932 8,14% > > > > > > Questions about this: > > > a) since our 90 BaaN never do start exceptional jobs and the nightjob > always > > > does the same routines I expected to see a stable BR after a few days. > Can > > > someone try to explain this, please ? > > > > Yes it should stabilize. > > > > > b) I also thought that the 3 BR parameters are running totals. But > pagreads > > > sometimes lowers. Maybe this is the answers to my first question, but is > > > this a normal behaviour? Why is this the only that behaves in this way? > > > > I see it but I don't believe it. Is it possible that the pagreads is > > wrapping around INT_MAX and back to zero? Considering the number of > complaints > > I see in CDI about Baan queries doing sequential scans, I think > > that more likely. Try taking a snapshot hourly and see what happens and > > when. > > > > > Thanks > > > > > > -- > > > *** remove spam twice when replying *** > > > > Almost noone will despammify you email address here. > > > > Art S. Kagel