Number and size of chunks

Hi, I have an instance with 528 chunks running on IDS 11.70 The server is a HP-UX 11.31 with KAIO enabled and with raw devices and lvols. The disk are SAN, aprox 25 LUNs in a combination of 30, 60 and 90 GB in size. All chunks are 2 GB in size. The CLEANERS parameter is 120. I would like to know if it is advisable to increase the size of each chunk and reorder the locations of the tables, so that the number of chunks does not increase more. I am concerned that in the short term the number of CLEANERS is not enough for the workload due to a greater number of chunks. Please let me any clues according your experience I'm sending some excerpts of relevant onstat commands. The instance is up for 22 days. Thanks in advance. onstat -F Fg Writes LRU Writes Chunk Writes 2 119362624 116202847 ... onstat -R ... 1019 m 2.0% 40 29 1118564 dirty, 999995 queued, 1000000 total, 1048576 hash buckets, 2048 buffer size start clean at 4.000% (of pair total) dirty, or 78 buffs dirty, stop at 2.000% onstat -g ckp AUTO_CKPTS=On RTO_SERVER_RESTART=Off Critical Sections Physical Log Logical Log Clock Total Flush Block # Ckpt Wait Long # Dirty Dskflu Total Avg Total Avg Interval Time Trigger LSN Time Time Time Waits Time Time Time Buffers /Sec Pages /Sec Pages /Sec 763889 16:07:51 CKPTINTVL 378244:0x10b10018 0.9 0.6 0.0 6 0.0 0.1 0.2 29529 29529 12493 40 15104 49 763890 16:12:54 CKPTINTVL 378244:0x12e03188 1.0 0.6 0.0 9 0.0 0.1 0.1 18137 18137 9324 30 8957 29 763891 16:17:58 CKPTINTVL 378244:0x157b5018 0.9 0.6 0.0 7 0.0 0.1 0.1 28167 28167 9280 30 10700 35 763892 16:23:02 CKPTINTVL 378245:0x790f4 0.7 0.5 0.0 4 0.0 0.0 0.1 28912 28912 11118 36 12194 40 763893 16:28:06 CKPTINTVL 378245:0x371008c 1.7 1.1 0.0 2 0.0 0.2 0.2 28894 26465 11947 39 13988 46 763894 16:33:10 CKPTINTVL 378245:0x6bf602c 0.7 0.5 0.0 6 0.0 0.1 0.1 25604 25604 11603 38 13546 44 763895 16:33:11 *Admin 378245:0x6bfc018 0.4 0.0 0.0 10 0.0 0.2 0.4 200 200 262 262 6 6 763896 16:36:58 *Admin 378245:0xa054018 0.4 0.1 0.0 7 0.0 0.3 0.4 1947 1947 8742 38 13400 59 763897 16:42:17 CKPTINTVL 378245:0xcdfc060 0.9 0.5 0.0 4 0.0 0.1 0.1 31085 31085 9962 31 11695 36 763898 16:47:21 CKPTINTVL 378245:0xfca7344 0.8 0.5 0.0 4 0.0 0.1 0.1 28109 28109 10823 35 11965 39 763899 16:52:25 CKPTINTVL 378245:0x126cb018 1.0 0.7 0.0 6 0.0 0.1 0.1 30474 30474 9389 30 10840 35 763900 16:57:29 CKPTINTVL 378245:0x16d29018 0.7 0.5 0.0 9 0.0 0.1 0.1 26153 26153 12025 39 18060 59 763901 17:02:35 CKPTINTVL 378246:0x3b14170 1.0 0.7 0.0 3 0.0 0.1 0.2 27028 27028 11992 39 21692 71 763902 17:07:41 CKPTINTVL 378246:0x70ad248 1.7 1.1 0.0 12 0.0 0.2 0.3 29023 27345 13220 43 13739 44 763903 17:12:47 CKPTINTVL 378246:0xb03f018 1.2 0.9 0.0 5 0.0 0.1 0.1 29566 29566 13366 43 16287 53 763904 17:17:53 CKPTINTVL 378246:0xd5595d8 1.9 1.4 0.0 5 0.0 0.1 0.1 31712 22114 8874 28 9505 30 763905 17:22:58 CKPTINTVL 378246:0xfd5a5a4 1.0 0.6 0.0 8 0.0 0.1 0.2 25124 25124 11273 36 10266 33 763906 17:28:05 CKPTINTVL 378246:0x12bad578 0.8 0.5 0.0 8 0.0 0.1 0.1 24247 24247 9316 30 11881 38 763907 17:33:10 CKPTINTVL 378246:0x158f9018 1.2 0.8 0.0 4 0.0 0.0 0.1 31988 31988 10187 33 11602 38 763908 17:38:15 CKPTINTVL 378247:0x1495018 1.4 1.0 0.0 4 0.0 0.1 0.2 30380 29107 10193 33 16961 55 Max Plog Max Llog Max Dskflush Avg Dskflush Avg Dirty Blocked pages/sec pages/sec Time pages/sec pages/sec Time 5013 16099 3 15474 3 0

Hi Roger, It would good to understand your wider concerns on system performance, for example are you concerned that a longer checkpoint would adversely affect your applications? There is a limit in Informix 11.70 of 32766 chunks (https://www.ibm.com/support/knowledgecenter/en/SSGU8G_11.70.0/com.ibm.adref.doc /ids_adr_0722.htm) so 528 chunks need not be a concern in itself. There is no requirement to keep chunks all the same size and, if you were to move to using storagepools, the engine itself can create some quite large chunks based on the size of a new chunk being 10% of the size of the whole dbspace. In my opinion if you are running 11.70 you do not need to be too concerned about long checkpoints (although watch the use of your physical log). The engine only performs a blocking checkpoint in a limited number of scenarios (start of L0 backup, block for external backup, maybe others). With this in mind you could, if anything, reduce the number of cleaners rather than be concerned that there are too few. If you monitor your checkpoint activity with "onstat -F" presumably you'd see very few of these active towards the end of the checkpoint. I don't see that you need to start moving tables around, at least not for this reason alone. If you were to, I'd give consideration to introducing larger page sizes where appropriate. HTH. Ben.

Thanks Benjamin