about chunks and raw partition

Support Informix — Wed, 16 Jul 2003 09:46:30 — source: IIUG Forums & Mailing Lists

Hello, (first, sorry for the html tags, I can't remove them) We have a little DB with a bizar chunk configuration. And I'm wondering why should my predecessors splitting up unix partitions in several chunks ? Unix partition /dev/rdblv03 (1Giga) is added as 1 big chunk in DBspace v2_dbs. This DBspace contains a test database "v04test". Unix partition /dev/rdblv01 (1Giga) is added as 5 chunks (each 100 or 200 Mega) in rootdbs and v1_dbs. The DBspace v1_dbs contains the production database "v04" splitted over 4 chunks. Do we have any benefits for DB "v04" spread over 4 chunks, but in the same raw device ? Would this allow more kio threads and so advantage the io performances ? What is the best we can do ? Next week we will dbexport-dbimport the data, so we can change things if necessary! Our system is an AIX 4.3 single cpu and local disk. Thanks a lot ? Yves Here is the result of the onstat -d Informix Dynamic Server Version 7.30.UC7 -- On-Line -- Up 6 days 22:08:50 -- 127984 Kbytes Dbspaces address number flags fchunk nchunks flags owner name 4003e13c 1 1 1 1 N informix rootdbs 4003e948 2 1 2 4 N informix v1_dbs 4003ea04 3 2001 4 1 N T informix tempdbs 4003eac0 4 1 7 1 N informix v2_dbs 4 active, 2047 maximum Chunks address chk/dbs offset size free bpages flags pathname 4003e1f8 1 1 0 50000 18961 PO- /dev/rdblv01 4003e420 2 2 50000 50000 14 PO- /dev/rdblv01 4003e4fc 3 2 100000 50000 50 PO- /dev/rdblv01 4003e5d8 4 3 0 250000 249947 PO- /dev/rdblv02 4003e6b4 5 2 150000 75000 705 PO- /dev/rdblv01 4003e790 6 2 225000 25000 14181 PO- /dev/rdblv01 4003e86c 7 4 0 250000 117195 PO- /dev/rdblv03 7 active, 2047 maximum ------_=_NextPart_001_01C34B7E.ECA71040 Content-Type: text/html; charset="iso-8859-1" Content-Transfer-Encoding: quoted-printable <!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 3.2//EN"> <HTML> <HEAD> <META HTTP-EQUIV=3D"Content-Type" CONTENT=3D"text/html; = charset=3Diso-8859-1"> <META NAME=3D"Generator" CONTENT=3D"MS Exchange Server version = 5.5.2653.12"> <TITLE>about chunks and raw partition </TITLE> </HEAD> <BODY> Hello, (first, sorry for the html tags, I can't remove = them) We have a little DB with a bizar chunk = configuration. And I'm wondering why should my predecessors = splitting up unix partitions in several chunks ? Unix partition /dev/rdblv03 (1Giga) is added as 1 big = chunk in DBspace v2_dbs. This DBspace contains a test database = "v04test". Unix partition /dev/rdblv01 (1Giga) is added as 5 = chunks (each 100 or 200 Mega) in rootdbs and v1_dbs. The DBspace v1_dbs contains the production database = "v04" splitted over 4 chunks. Do we have any benefits for DB "v04" spread = over 4 chunks, but in the same raw device ? Would this allow more kio threads and so advantage = the io performances ? What is the best we can do ? Next week we will dbexport-dbimport the data, so we = can change things if necessary! Our system is an AIX 4.3 single cpu and local = disk. Thanks a lot ? Yves Here is the result of the onstat -d Informix Dynamic Server Version 7.30.UC7   = -- On-Line -- Up 6 days 22:08:50 -- 127984 Kbytes Dbspaces address  number   = flags    fchunk   nchunks  = flags    owner    name 4003e13c 1        = 1        = 1        = 1        = N        informix rootdbs 4003e948 2        = 1        = 2        = 4        = N        informix v1_dbs 4003ea04 3        = 2001     = 4        = 1        N = T      informix tempdbs 4003eac0 4        = 1        = 7        = 1        = N        informix v2_dbs  4 active, 2047 maximum Chunks address  chk/dbs offset   = size     free     = bpages   flags pathname 4003e1f8 1   1   = 0        50000    = 18961           &= nbsp; PO-   /dev/rdblv01 4003e420 2   2   = 50000    50000    = 14           &nbs= p;    PO-   /dev/rdblv01 4003e4fc 3   2   = 100000   50000    = 50           &nbs= p;    PO-   /dev/rdblv01 4003e5d8 4   3   = 0        250000   = 249947           = PO-   /dev/rdblv02 4003e6b4 5   2   = 150000   75000    = 705           &nb= sp;   PO-   /dev/rdblv01 4003e790 6   2   = 225000   25000    = 14181           &= nbsp; PO-   /dev/rdblv01 4003e86c 7   4   = 0        250000   = 117195           = PO-   /dev/rdblv03  7 active, 2047 maximum </BODY> </HTML>@

↪ replying to Support Informix

Jack Parker — Wed, 16 Jul 2003 13:04:43 — source: IIUG Forums & Mailing Lists

Happens sometimes, you keep a spare disk lying around, then you need a little disk for this database, or a little disk for that one, so you grab a little. No performance benefit at all. It's kind of sloppy, but not too harmful - if it ain't broke....., but next time you reorg you could drop and re-add as one chunk - just to be tidy. cheers j. ----- Original Message ----- From: "Support Inf...." <supinformix@dofi.fgov.be> To: <ids@iiug.org> Sent: Wednesday, July 16, 2003 5:46 AM Subject: about chunks and raw partition [1560] > Hello, > > (first, sorry for the html tags, I can't remove them) > > We have a little DB with a bizar chunk configuration. > And I'm wondering why should my predecessors splitting up unix partitions in > several chunks ? > > Unix partition /dev/rdblv03 (1Giga) is added as 1 big chunk in DBspace > v2_dbs. > This DBspace contains a test database "v04test". > > Unix partition /dev/rdblv01 (1Giga) is added as 5 chunks (each 100 or 200 > Mega) in rootdbs and v1_dbs. > The DBspace v1_dbs contains the production database "v04" splitted over 4 > chunks. > > Do we have any benefits for DB "v04" spread over 4 chunks, but in the same > raw device ? > Would this allow more kio threads and so advantage the io performances ? > What is the best we can do ? > Next week we will dbexport-dbimport the data, so we can change things if > necessary! > > > Our system is an AIX 4.3 single cpu and local disk. > > Thanks a lot ? > > Yves > > > Here is the result of the onstat -d > Informix Dynamic Server Version 7.30.UC7 -- On-Line -- Up 6 days 22:08:50 > -- 127984 Kbytes > > Dbspaces > address number flags fchunk nchunks flags owner name > 4003e13c 1 1 1 1 N informix rootdbs > 4003e948 2 1 2 4 N informix v1_dbs > 4003ea04 3 2001 4 1 N T informix tempdbs > 4003eac0 4 1 7 1 N informix v2_dbs > 4 active, 2047 maximum > > Chunks > address chk/dbs offset size free bpages flags pathname > 4003e1f8 1 1 0 50000 18961 PO- /dev/rdblv01 > 4003e420 2 2 50000 50000 14 PO- /dev/rdblv01 > 4003e4fc 3 2 100000 50000 50 PO- /dev/rdblv01 > 4003e5d8 4 3 0 250000 249947 PO- /dev/rdblv02 > 4003e6b4 5 2 150000 75000 705 PO- /dev/rdblv01 > 4003e790 6 2 225000 25000 14181 PO- /dev/rdblv01 > 4003e86c 7 4 0 250000 117195 PO- /dev/rdblv03 > 7 active, 2047 maximum > > ------_=_NextPart_001_01C34B7E.ECA71040 > Content-Type: text/html; > charset="iso-8859-1" > Content-Transfer-Encoding: quoted-printable > > <!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 3.2//EN"> > <HTML> > <HEAD> > <META HTTP-EQUIV=3D"Content-Type" CONTENT=3D"text/html; = > charset=3Diso-8859-1"> > <META NAME=3D"Generator" CONTENT=3D"MS Exchange Server version = > 5.5.2653.12"> > <TITLE>about chunks and raw partition </TITLE> > </HEAD> > <BODY> > > Hello, > > > (first, sorry for the html tags, I can't remove = > them) > > > We have a little DB with a bizar chunk = > configuration. > And I'm wondering why should my predecessors = > splitting up unix partitions in several chunks ? > > > Unix partition /dev/rdblv03 (1Giga) is added as 1 big = > chunk in DBspace v2_dbs. > This DBspace contains a test database = > "v04test". > > > Unix partition /dev/rdblv01 (1Giga) is added as 5 = > chunks (each 100 or 200 Mega) in rootdbs and v1_dbs. > The DBspace v1_dbs contains the production database = > "v04" splitted over 4 chunks. > > > Do we have any benefits for DB "v04" spread = > over 4 chunks, but in the same raw device ? > Would this allow more kio threads and so advantage = > the io performances ? > What is the best we can do ? > Next week we will dbexport-dbimport the data, so we = > can change things if necessary! > > > > Our system is an AIX 4.3 single cpu and local = > disk. > > > Thanks a lot ? > > > Yves > > > > Here is the result of the onstat -d > Informix Dynamic Server Version 7.30.UC7   = > -- On-Line -- Up 6 days 22:08:50 -- 127984 Kbytes > > > Dbspaces > address  number   = > flags    fchunk   nchunks  = > flags    owner    name > 4003e13c 1        = > 1        = > 1        = > 1        = > N        informix rootdbs > 4003e948 2        = > 1        = > 2        = > 4        = > N        informix v1_dbs > 4003ea04 3        = > 2001     = > 4        = > 1        N = > T      informix tempdbs > 4003eac0 4        = > 1        = > 7        = > 1        = > N        informix v2_dbs >  4 active, 2047 maximum > > > Chunks > address  chk/dbs offset   = > size     free     = > bpages   flags pathname > 4003e1f8 1   1   = > 0        50000    = > 18961           &= > nbsp; PO-   /dev/rdblv01 > 4003e420 2   2   = > 50000    50000    = > 14           &nbs= > p;    PO-   /dev/rdblv01 > 4003e4fc 3   2   = > 100000   50000    = > 50           &nbs= > p;    PO-   /dev/rdblv01 > 4003e5d8 4   3   = > 0        250000   = > 249947 &nbsp

↪ replying to Support Informix

Abraham Kirubakaran — Fri, 18 Jul 2003 19:37:30 — source: IIUG Forums & Mailing Lists

I would suggest you to put the "rootdbs","v1_dbs&v2_dbs" and "tempdbs" in the raw devices by themselves. Also, the informix engine multithreads on the dbspaces, so just make the dbspaces 2Gb chunks and keep the critical ones seperate. -Abraham --- "Support Inf...." <supinformix@dofi.fgov.be> wrote: > Hello, > > (first, sorry for the html tags, I can't remove > them) > > We have a little DB with a bizar chunk > configuration. > And I'm wondering why should my predecessors > splitting up unix partitions in > several chunks ? > > Unix partition /dev/rdblv03 (1Giga) is added as 1 > big chunk in DBspace > v2_dbs. > This DBspace contains a test database "v04test". > > Unix partition /dev/rdblv01 (1Giga) is added as 5 > chunks (each 100 or 200 > Mega) in rootdbs and v1_dbs. > The DBspace v1_dbs contains the production database > "v04" splitted over 4 > chunks. > > Do we have any benefits for DB "v04" spread over 4 > chunks, but in the same > raw device ? > Would this allow more kio threads and so advantage > the io performances ? > What is the best we can do ? > Next week we will dbexport-dbimport the data, so we > can change things if > necessary! > > > Our system is an AIX 4.3 single cpu and local disk. > > Thanks a lot ? > > Yves > > > Here is the result of the onstat -d > Informix Dynamic Server Version 7.30.UC7 -- > On-Line -- Up 6 days 22:08:50 > -- 127984 Kbytes > > Dbspaces > address number flags fchunk nchunks flags > owner name > 4003e13c 1 1 1 1 N > informix rootdbs > 4003e948 2 1 2 4 N > informix v1_dbs > 4003ea04 3 2001 4 1 N T > informix tempdbs > 4003eac0 4 1 7 1 N > informix v2_dbs > 4 active, 2047 maximum > > Chunks > address chk/dbs offset size free bpages > flags pathname > 4003e1f8 1 1 0 50000 18961 > PO- /dev/rdblv01 > 4003e420 2 2 50000 50000 14 > PO- /dev/rdblv01 > 4003e4fc 3 2 100000 50000 50 > PO- /dev/rdblv01 > 4003e5d8 4 3 0 250000 249947 > PO- /dev/rdblv02 > 4003e6b4 5 2 150000 75000 705 > PO- /dev/rdblv01 > 4003e790 6 2 225000 25000 14181 > PO- /dev/rdblv01 > 4003e86c 7 4 0 250000 117195 > PO- /dev/rdblv03 > 7 active, 2047 maximum > > ------_=_NextPart_001_01C34B7E.ECA71040 > Content-Type: text/html; > charset="iso-8859-1" > Content-Transfer-Encoding: quoted-printable > > <!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 3.2//EN"> > <HTML> > <HEAD> > <META HTTP-EQUIV=3D"Content-Type" > CONTENT=3D"text/html; = > charset=3Diso-8859-1"> > <META NAME=3D"Generator" CONTENT=3D"MS Exchange > Server version = > 5.5.2653.12"> > <TITLE>about chunks and raw partition </TITLE> > </HEAD> > <BODY> > > Hello, > > > (first, sorry for the html tags, I > can't remove = > them) > > > We have a little DB with a bizar > chunk = > configuration. > And I'm wondering why should my > predecessors = > splitting up unix partitions in several chunks > ? > > > Unix partition /dev/rdblv03 > (1Giga) is added as 1 big = > chunk in DBspace v2_dbs. > This DBspace contains a test > database = > "v04test". > > > Unix partition /dev/rdblv01 > (1Giga) is added as 5 = > chunks (each 100 or 200 Mega) in rootdbs and > v1_dbs. > The DBspace v1_dbs contains the > production database = > "v04" splitted over 4 chunks. > > > Do we have any benefits for DB > "v04" spread = > over 4 chunks, but in the same raw device ? > Would this allow more kio threads > and so advantage = > the io performances ? > What is the best we can do > ? > Next week we will > dbexport-dbimport the data, so we = > can change things if necessary! > > > > Our system is an AIX 4.3 single > cpu and local = > disk. > > > Thanks a lot ? > > > Yves > > > > Here is the result of the onstat > -d > Informix Dynamic Server Version > 7.30.UC7   = > -- On-Line -- Up 6 days 22:08:50 -- 127984 > Kbytes > > > Dbspaces > address  number   > = > flags    fchunk   > nchunks  = > flags    owner    > name > 4003e13c > 1        = > 1        = > 1        = > 1        = > N        informix > rootdbs > 4003e948 > 2        = > 1        = > 2        = > 4        = > N        informix > v1_dbs > 4003ea04 > 3        = > 2001     = > 4        = > 1        N = > T      informix > tempdbs > 4003eac0 > 4        = > 1        = > 7        = > 1        = > N        informix > v2_dbs >  4 active, 2047 > maximum > === message truncated === __________________________________ Do you Yahoo!? SBC Yahoo! DSL - Now only $29.95 per month! http://sbc.yahoo.com

Related threads

Glossary

Your privacy choices