What part of zOS is the OS 4389

Anne & Lynn Wheeler

ref:

this brings to my mind somebody's line that goes something like "it isn't done when there is no more to add, it is done when there is no more to take away"

applied to micro-kernel efforts ... with somebody's related observation that maintaining a KISS implementation can actually be significantly more difficult than doing a complex implementation.

however, there is a corollary about KISS implementation being applicable to the situation. recent posts mentioning how simple handling of FINWAIT processing in tcp session close ... made some implicit buttumptions about the environment that were violated with HTTP use of TCP sessions. protection against DOS protection against DOS

the mention of the spool file system rewrite in the previous posts MVS???

addressed a problem similar to what showed up with FINWAIT handling on heavily loaded webservers in the mid-90s. native vm spool processing had a linear list of all spool files ... and all spool file operations involved searching the linear list. this had non-linear increase in overhead as systems scaled. this is also similar to the original cp67 kernel storage management that used a single linear list ... before subpool logic was introduced to cp67 kernel in the early 70s.

in any case, my pascal-vs spool file rewrite introduced both a hash table and a tree structure for managing spool files.