I'm pretty sure the original motivation behind VSAM was to provide a
more efficient keyed-access alternative to an earlier ISAM (Indexed
Sequential Access Method) keyed-access method --ISAM had abysmal
performance when the number of inserts became large. But, just as a
clarification, while it is true that VSAM under IBM Operating systems
includes support for keyed-access data sets in the forms of KSDS (Keyed
Sequential) and RRDS (Relative Record), VSAM also includes support for
non-keyed-access data set types: ESDS (Entry-Sequenced) and LDS
(Linear).
The LDS data sets on z/OS do not have any organizational structure
imposed upon them by VSAM other than being a stream of consecutive bytes
physically organized in 4KiB, pageable chunks; and DB2 uses these LDS
data sets for its table spaces. So, it is incorrect to think of DB2 as
somehow using or building upon VSAM keyed-access techniques to locate
table rows. DB2 has it own table structures, indexing organization, and
key search techniques built on top of the very rudimentary Linear Data
Set structure.
DB2 required greater flexibility and control to achieve its performance
and data consistency goals than was possible with VSAM keyed-access
support. A very cursory look at "Experiences Installing Oracle
Database 10g on z/OS" would suggest Oracle on z/OS makes use of VSAM LDS
for table storage as well.
Joel C Ewing
On 04/21/2014 04:52 AM, Bernd Oppolzer wrote:
> Ok, I was not aware of that fact. Thank you.
>
> It is indeed a clever decision to make such "Access Method Services"
> part of the operating system, because this way a standard method is
> defined
> and all other languages or systems can use it. For example DB2, which
> uses VSAM as access method in several respects. (Even ORACLE is
> using VSAM, when it runs on z-Hardware, IIRC ... and so ORACLE
> benefits from VSAM improvement done to support DB2).
>
> LE comes to mind; such services, available to all programming languages,
> should also be part of the operating system.
>
> The German Telefunken mainframe TR440's operating system BS3 also
> had keyed file access services (different types) and a common run time
> environment
> for all languages (ALGOL, FORTRAN, PASCAL, PL/1, COBOL, BCPL, ...).
> And it had (in the 1970s) a common method of storing
> diagnose information at compile time, helping the runtime environment
> to write helpful dump information in the case of an abend. The variables
> were all shown in source language syntax; there was no need to look
> at hex storage dumps. This whole technique was part of the operating
> system, too. (Even interactive symbolic debugging ... using teletypes and
> very basic display terminals ... was possible).
>
> As I said, the opinions what a operating system should do, changed
> over time, even at IBM. There was never such support for programmers
> as I outlined it above in IBM systems, although in the last ten years
> or so,
> LE developed in this direction, too.
>
> As you can see, I am no advocat of Windows etc.; the support for
> programmers there is non-existent ... if you need it, you have to buy
> expensive compiler suites from M$ ... Linux and Unix is different, of
> course.
>
> It is a big problem in Windows (and DOS etc. in previous times), that the
> compilers all had different call sequences and parameter passing
> mechanisms,
> so that it was not possible to mix languages and/or compilers. Such
> things
> as common linkage conventions are very helpful (although there are some
> problems at the z/Arch, too).
>
> I have a vision how an ideal operating system should look like from the
> programmers point of view ... and that vision is very much inspired by my
> old experience of the 1970s operating system BS3. Sorry.
>
> Kind regards
>
> Bernd
>
>
>
> Am 21.04.2014 07:14, schrieb Timothy Sipples:
>> Bernd Oppolzer asks:
>>> BTW: is VSAM part of the operating system?
>> Yes. VSAM is certainly a standard, included, essential, no additional
>> charge feature within the base z/OS and z/VSE operating systems.
>>
>> IBM first introduced VSAM (in a more basic form of course) back in the
>> 1970s for DOS/VS, OS/VS1, and OS/VS2 ("Access Method Services"). I don't
>> think VSAM was ever separately licensed, but someone can correct me
>> if I'm
>> mistaken on that point.
>>
>> z/VM's CMS and MUSIC/SP also support VSAM, also in base. Base z/TPF does
>> also, though it collaborates with z/OS to get the job done. Hitachi's
>> and
>> Fujitsu's Japanese mainframe operating systems (VOS3, MSP, XSP) also
>> support VSAM -- not a great surprise given their heritages.
>>
>> Perhaps you're thinking of Transactional VSAM (DFSMStvs)? Transactional
>> VSAM is an optional addition available for z/OS. (VSAM RLS is part of
>> base
>> z/OS.)
>>
>> --------------------------------------------------------------------------------------------------------
>>
>> Timothy Sipples
>> IT Architect Executive, zEnterprise Industry Solutions, AP/GCG/MEA
>> E-Mail: [email protected]
--
Joel C. Ewing, Bentonville, AR [email protected]
----------------------------------------------------------------------
For IBM-MAIN subscribe / signoff / archive access instructions,
send email to [email protected] with the message: INFO IBM-MAIN
