Thanks for the question, YULIN.
Asked: December 07, 2020 - 12:48 pm UTC
Last updated: December 08, 2020 - 7:39 am UTC
Version: 11.2.0.4
Viewed 1000+ times
You Asked
Why there would always be many non-zero long_waits values in the V$BACKUP_ASYNC_IO, which indicates waiting the O/S blocking I/O completed？
Notes, here what the difference between the blocking I/O and synchronous I/O ？
IO_COUNT READY SHORT_WAITS LONG_WAITS FILENAME TO_CHAR(CLOSE_TIME, EPS
---------- ---------- ----------- ---------- -------------------------------------------------------- ------------------- ----------
27 26 0 1 +DATA/cpemsdb/data_d-cpemsdb_ts-h_tzxx_s1_fno-389 11/27/2020 23:46:53 394202
27 25 0 2 +DATA/cpemsdb/data_d-cpemsdb_ts-data_m_fno-198 11/27/2020 23:46:51 389805
7 7 0 0 +DATA/cpemsdb/data_d-cpemsdb_ts-bbxx_fno-38 11/27/2020 23:46:33 2097152
7682 7104 0 578 +DATA/cpemsdb/data_d-cpemsdb_ts-idx_s_fno-676 11/27/2020 23:46:25 4145188
7682 6905 0 777 +DATA/cpemsdb/data_d-cpemsdb_ts-idx_p_fno-612 11/27/2020 23:46:21 4141990
7682 7165 0 517 +DATA/cpemsdb/data_d-cpemsdb_ts-idx_e_fno-515 11/27/2020 23:42:24 4266524
7682 7166 0 516 +DATA/cpemsdb/data_d-cpemsdb_ts-idx_a_fno-401 11/27/2020 23:42:18 4265394
7682 7167 0 515 +DATA/cpemsdb/data_d-cpemsdb_ts-data_p_fno-316 11/27/2020 23:41:51 4276720
7682 7166 0 516 +DATA/cpemsdb/data_d-cpemsdb_ts-data_out_fno-260 11/27/2020 23:41:25 4288106
7682 7166 0 516 +DATA/cpemsdb/data_d-cpemsdb_ts-data_man_fno-203 11/27/2020 23:40:57 4301850
7682 7047 0 635 +DATA/cpemsdb/data_d-cpemsdb_ts-data_a_fno-43 11/27/2020 23:40:14 4322052
82537 76975 0 5562 11/27/2020 22:22:57 37828972
7682 7137 0 545 +DATA/cpemsdb/data_d-cpemsdb_ts-data_out_fno-781 11/27/2020 22:22:57 11422785
5122 4773 0 349 +DATA/cpemsdb/data_d-cpemsdb_ts-idx_arc_fno-439 11/27/2020 22:16:12 8910721
4098 3821 0 277 +DATA/cpemsdb/data_d-cpemsdb_ts-idx_e_fno-504 11/27/2020 22:10:47 8271483
4098 3821 0 277 +DATA/cpemsdb/data_d-cpemsdb_ts-data_arc_fno-115 11/27/2020 22:10:45 8263525
7682 6859 0 823 +DATA/cpemsdb/data_d-cpemsdb_ts-idx_arc_fno-780 11/27/2020 21:44:07 28658590
82537 76630 0 5907 11/27/2020 21:44:07 44843779
5122 4746 0 376 +DATA/cpemsdb/data_d-cpemsdb_ts-idx_arc_fno-438 11/27/2020 21:40:36 23650701
4098 3812 0 286 +DATA/cpemsdb/data_d-cpemsdb_ts-idx_e_fno-498 11/27/2020 21:39:24 20773723
4098 3813 0 285 +DATA/cpemsdb/data_d-cpemsdb_ts-data_arc_fno-97 11/27/2020 21:39:22 20698638
27 25 0 2 +DATA/cpemsdb/data_d-cpemsdb_ts-idx_yxbz_fno-689 11/27/2020 21:36:17 14979657
27 26 0 1 +DATA/cpemsdb/data_d-cpemsdb_ts-data_ics_fno-193 11/27/2020 21:36:15 20971520
27 25 0 2 +DATA/cpemsdb/data_d-cpemsdb_ts-hgqjd_fno-384 11/27/2020 21:36:15 20971520
7682 7168 0 514 +DATA/cpemsdb/data_d-cpemsdb_ts-idx_s_fno-674 11/27/2020 21:36:11 5206442
7682 7168 0 514 +DATA/cpemsdb/data_d-cpemsdb_ts-idx_out_fno-609 11/27/2020 21:36:10 5191338
7682 7169 0 513 +DATA/cpemsdb/data_d-cpemsdb_ts-idx_e_fno-501 11/27/2020 21:36:10 5175491
7682 7168 0 514 +DATA/cpemsdb/data_d-cpemsdb_ts-idx_a_fno-399 11/27/2020 21:36:10 5157261
7682 7169 0 513 +DATA/cpemsdb/data_d-cpemsdb_ts-data_p_fno-314 11/27/2020 21:36:06 5139980
7682 7169 0 513 +DATA/cpemsdb/data_d-cpemsdb_ts-data_out_fno-258 11/27/2020 21:36:02 5125259
7682 7168 0 514 +DATA/cpemsdb/data_d-cpemsdb_ts-data_man_fno-201 11/27/2020 21:35:57 5113056
82542 76562 0 5980 11/27/2020 21:35:52 44881392
7682 6787 0 895 +DATA/cpemsdb/data_d-cpemsdb_ts-idx_arc_fno-779 11/27/2020 21:35:52 27531842
7682 7168 0 514 +DATA/cpemsdb/data_d-cpemsdb_ts-arc_r_data_fno-37 11/27/2020 21:35:49 5096876
5122 4753 0 369 +DATA/cpemsdb/data_d-cpemsdb_ts-idx_arc_fno-437 11/27/2020 21:33:46 20930640
4098 3811 0 287 +DATA/cpemsdb/data_d-cpemsdb_ts-idx_e_fno-496 11/27/2020 21:32:59 18046081
4098 3812 0 286 +DATA/cpemsdb/data_d-cpemsdb_ts-data_arc_fno-94 11/27/2020 21:32:57 17877075
27 25 0 2 +DATA/cpemsdb/data_d-cpemsdb_ts-idx_y_fno-688 11/27/2020 21:25:50 8738133
27 24 0 3 +DATA/cpemsdb/data_d-cpemsdb_ts-goldengate_fno-383 11/27/2020 21:25:49 9532509
27 25 0 2 +DATA/cpemsdb/data_d-cpemsdb_ts-data_ccom_fno-176 11/27/2020 21:25:47 10485760
7682 7166 0 516 +DATA/cpemsdb/data_d-cpemsdb_ts-idx_out_fno-608 11/27/2020 21:25:38 5028451
7682 7165 0 517 +DATA/cpemsdb/data_d-cpemsdb_ts-idx_s_fno-673 11/27/2020 21:25:38 5048151
7682 7165 0 517 +DATA/cpemsdb/data_d-cpemsdb_ts-idx_e_fno-500 11/27/2020 21:25:37 5003457
7682 7166 0 516 +DATA/cpemsdb/data_d-cpemsdb_ts-idx_a_fno-398 11/27/2020 21:25:37 4974866
7682 7166 0 516 +DATA/cpemsdb/data_d-cpemsdb_ts-data_p_fno-313 11/27/2020 21:25:32 4954207
7682 7165 0 517 +DATA/cpemsdb/data_d-cpemsdb_ts-data_out_fno-257 11/27/2020 21:25:28 4933720
7682 7166 0 516 +DATA/cpemsdb/data_d-cpemsdb_ts-data_man_fno-200 11/27/2020 21:25:23 4915650
7682 7167 0 515 +DATA/cpemsdb/data_d-cpemsdb_ts-arc_r_data_fno-36 11/27/2020 21:25:10 4895479
52 46 0 6 +DATA/cpemsdb/data_d-cpemsdb_ts-veritas_apm_fno-728 11/27/2020 21:17:45 5991863
27 26 0 1 +DATA/cpemsdb/data_d-cpemsdb_ts-idx_tzxx_s3_fno-686 11/27/2020 21:17:34 4766255
7 7 0 0 +DATA/cpemsdb/data_d-cpemsdb_ts-h_tzxx_s3_fno-394 11/27/2020 21:17:21 3495253
7682 7165 0 517 +DATA/cpemsdb/data_d-cpemsdb_ts-idx_s_fno-672 11/27/2020 21:17:16 4987189
7682 7164 0 518 +DATA/cpemsdb/data_d-cpemsdb_ts-idx_out_fno-607 11/27/2020 21:17:13 4975634
7682 7166 0 516 +DATA/cpemsdb/data_d-cpemsdb_ts-idx_e_fno-499 11/27/2020 21:17:11 4964132
7682 7164 0 518 +DATA/cpemsdb/data_d-cpemsdb_ts-idx_a_fno-397 11/27/2020 21:17:08 4949640
7682 7165 0 517 +DATA/cpemsdb/data_d-cpemsdb_ts-data_p_fno-312 11/27/2020 21:16:57 4943563
7682 7166 0 516 +DATA/cpemsdb/data_d-cpemsdb_ts-data_out_fno-256 11/27/2020 21:16:41 4941288
7682 7165 0 517 +DATA/cpemsdb/data_d-cpemsdb_ts-data_man_fno-199 11/27/2020 21:16:24 4943563
7682 7165 0 517 +DATA/cpemsdb/data_d-cpemsdb_ts-arc_r_data_fno-35 11/27/2020 21:15:53 4946599
82542 76436 0 6106 11/27/2020 19:51:17 32569209
7682 6928 0 754 +DATA/cpemsdb/data_d-cpemsdb_ts-idx_out_fno-778 11/27/2020 19:51:17 18449172
7682 6966 0 716 +DATA/cpemsdb/data_d-cpemsdb_ts-idx_out_fno-777 11/27/2020 19:49:28 19311903
82542 76530 0 6012 11/27/2020 19:49:28 32480564
5122 4761 0 361 +DATA/cpemsdb/data_d-cpemsdb_ts-idx_arc_fno-436 11/27/2020 19:47:09 14569089
5122 4760 0 362 +DATA/cpemsdb/data_d-cpemsdb_ts-idx_arc_fno-432 11/27/2020 19:45:39 15101854
4098 3814 0 284 +DATA/cpemsdb/data_d-cpemsdb_ts-idx_d_fno-488 11/27/2020 19:45:28 12678870
4098 3815 0 283 +DATA/cpemsdb/data_d-cpemsdb_ts-data_a_fno-88 11/27/2020 19:45:26 12576771
4098 3821 0 277 +DATA/cpemsdb/data_d-cpemsdb_ts-idx_arc_fno-472 11/27/2020 19:43:57 13144506
4098 3818 0 280 +DATA/cpemsdb/data_d-cpemsdb_ts-data_a_fno-87 11/27/2020 19:43:56 13114404
82542 76571 0 5971 11/27/2020 19:34:44 34954087
7682 6911 0 771 +DATA/cpemsdb/data_d-cpemsdb_ts-idx_man_fno-775 11/27/2020 19:34:44 18826566
5122 4755 0 367 +DATA/cpemsdb/data_d-cpemsdb_ts-idx_arc_fno-431 11/27/2020 19:31:15 14648592
4098 3816 0 282 +DATA/cpemsdb/data_d-cpemsdb_ts-idx_arc_fno-471 11/27/2020 19:29:36 12956161
4098 3818 0 280 +DATA/cpemsdb/data_d-cpemsdb_ts-data_a_fno-84 11/27/2020 19:29:36 12707004
82589 76631 0 5958 11/27/2020 19:26:49 35341656
7554 7014 0 540 +DATA/cpemsdb/data_d-cpemsdb_ts-idx_out_fno-611 11/27/2020 19:26:49 10795973
5618 5230 0 388 +DATA/cpemsdb/data_d-cpemsdb_ts-idx_e_fno-516 11/27/2020 19:24:24 8581133
52 48 0 4 +DATA/cpemsdb/data_d-cpemsdb_ts-idx_pay_fno-661 11/27/2020 19:23:48 3883615
and Connor said...
You didn't give the TYPE column, but I'll make the assumption that
- they are mostly the OUTPUT type
- the null filenames are your aggregate type
For asynchronous I/O, you would generally expect to see some "long waits" no matter how good your storage is, because for almost any kind of storage, its faster to read it than it is to write it. For async I/O, we "want* to get some waiting because we're going to fire off I/O requests in parallel to get as much write performance out of the target device as we can. If long waits was always zero, it would indicate that we probably could be pushing *more* through. For example, as a metaphor,
a)- if there is *no* queue at any of checkout lanes at the supermarket, it means the supermarket can handle more load
b)- if there are massive queues at all of the checkout lands, we're overloaded - we need more or faster checkouts
c)- if there is a small queue at each checkout, we are pretty much at the optimal rate for our checkout.
From your data, you're typically seeing around 5% long waits (compared to total I/O) which puts you in (c) above. The metaphor is not entirely accurate, because RMAN is not going to push the storage subsystem harder than its capable, In particular, I'm *guessing* that EPS is the effective bytes per second column, and if so, then those rates are not spectacular...
Rating
(1 rating)
Is this answer out of date? If it is, please let us know via a Comment