On 01/29/2010 07:59 AM, Paul Gilmartin wrote: > On Thu, 28 Jan 2010 10:33:49 -0500, Thompson, Steve wrote: >> >> So when you did this call to CICS, it would recognize you would need to >> wait (whether you actually needed to or now, you will now) and then it >> would go dispatch another transaction that was ready to run. >> >> In effect, CICS was making virtual TCBs run using one real TCB, with >> each virtual TCB being a transaction (pseudo conversational or true >> conversational). >> > Why re-invent the wheel? But this feels more like newly inventing > the flat tire.
The obvious (?) answer here is that because of the additional constraints placed on CICS application transaction code, CICS was able to do its own internal CPU dispatching much more efficiently than using separate TCBs for all transactions and letting MVS manage everything. This was certainly true when CICS was created. Perhaps even more critical, having CICS manage things made it possible to share update access to VSAM datasets and manage other transaction resources in ways that would have been impractical or have involved more overhead with multiple system TCBs. 30+ years ago we had exactly the same issues with DOS/VS and its support for multitasking. In pre-CICS days, to support our own on-line system on limited hardware resources my predecessors had to develop what we called "mini-tasking" - a set of interfaces and macros to support multiple interactive transactions as a single system task in a single DOS partition -- a much simplified version of what CICS does. Those applications eventually evolved into CICS transactional applications, which thankfully proved much easier to migrate to MVS. > > I suppose there's a transaction manager that needs o be aware when > a transaction waits. "Virtual" seems to be a key word. It's a > pity there wasn't an implementation closer to the nucleus, so > applications could use a fuller repertoire of conventional MVS > multitasking. A hierarchy of virtual environments? SVC screening > of WAIT by the task manager? > > -- gil -- Joel C. Ewing, Fort Smith, AR jremoveccapsew...@acm.org ---------------------------------------------------------------------- For IBM-MAIN subscribe / signoff / archive access instructions, send email to lists...@bama.ua.edu with the message: GET IBM-MAIN INFO Search the archives at http://bama.ua.edu/archives/ibm-main.html