Differences in RSAM page structure 32 and 64 bit I

Greetings, all. With some delusions of grandeur and other aspirations, I'm more than a little curious: We all (or most of us) are familiar with the 24-byte page header and the (somewhat misnamed) time-stamp at the end of each page in an IDS server. (I'm not certain that latter time stamp is used anymore..) But I never found out for certain: For 64-bit IDS, especially with large chunks enabled, is the page-header structure the same as for 32-bit IDS? With the increased limit for pages in a chunk, as well as chunks in a server, I would think the page address would no longer fit into the neat 20+12 bit word that sits (or sat?) at the beginning of each page. In a chunk of 32,000 pages, the 12-bit offset is inadequate for the chunk-relative page number. Would someone please post the structure of the RSAM page header as it applies to IDS 11 and up? And if different for 32 and 64 bit IDS, please post both. Thanks much! -- Jacob S (In pursuit of undomesticated, semi-aquatic avians)

On 07/30/14 00:25, JACOB SALOMON wrote: > Greetings, all. > > With some delusions of grandeur and other aspirations, I'm more than a little > curious: > > We all (or most of us) are familiar with the 24-byte page header and the > (somewhat misnamed) time-stamp at the end of each page in an IDS server. (I'm > not certain that latter time stamp is used anymore..) But I never found out > for certain: For 64-bit IDS, especially with large chunks enabled, is the > page-header structure the same as for 32-bit IDS? With the increased limit for > pages in a chunk, as well as chunks in a server, I would think the page > address would no longer fit into the neat 20+12 bit word that sits (or sat?) > at the beginning of each page. In a chunk of 32,000 pages, the 12-bit offset > is inadequate for the chunk-relative page number. > > Would someone please post the structure of the RSAM page header as it applies > to IDS 11 and up? And if different for 32 and 64 bit IDS, please post both. > > Thanks much! > > -- Jacob S (In pursuit of undomesticated, semi-aquatic avians) Yup - the page header is the same for 32 and 64 bit instances, and is still 24 bytes. Pages can be either in old or big chunk format (there's a page flag to show the mode in use). In old format, things, unsurprisingly, work as usual, ie the page header is made up of page address, timestamp #1, no of slots, flags, etc. In big chink mode the page header is composed of page offset (32 bits), chunk number (16 bits), checksum (16 bits), etc. The checksum is used to verify pages consistency, much like timestamp #1 and #2 are used in old page format, except that now rather than comparing two values and complaining if they are found to be different, we work out a checksum value from the page header and timestamp #2 and compare that to the checksum. Things are further complicated by variable page size - in that we have to store the page size for the page somewhere, but that's a topic for another discussion. This stuff is nicely covered in the Internal Architecture course - just in case you are interested. -- Ciao, Marco ______________________________________________________________________________ Marco Greco /UK /IBM Standard disclaimers apply! Structured Query Scripting Language http://www.4glworks.com/sqsl.htm 4glworks http://www.4glworks.com Informix on Linux http://www.4glworks.com/ifmxlinux.htm