When Does Folklore Begin 4188
 |
ref:
it had support for the same api-semantics as offered by vms dlm ... but internally it kept-piggybacked state in more places so that it shorten recovery time (slightly more upfront work but shortened any fault-recovery time).
some of the vendors had fast commit or something similar in non-cluster environment ... but was disabled for any cluster operation. as part of the dlm work, i had also worked out the conventions for doing fast commit type operations in cluster environment. however, originally it was considered somewhat too advanced ... although I've had follow-up in the past couple years about implementations for doing fast commit type operation in cluster environment.
misc. past posts mentioning fast commit stuff:
other ha-cmp posts
and other posts on continuous availability, disaster survivability, and geographic survivability
I have recently made a correction to some illustrations of APL keyboards on my web site. It was difficult for me to find the...
note, some amount of the work also drew on my wife's experience doing a stint (long ago and far away) in POK responsible for mainframe loosely-coupled architecture (i.e. loosely-coupled is mainframe speek for cluster, dates back to 360 cluster operation in the 60s). while there, she developed peer-coupled shared data architecture
which didn't see much uptake at the time ... but eventually used by IMS hot-standby ... and since has evolved into (mainframe) parallel sysplex.
a parallel sysplex URL:
misc. past posts mentioning parallel sysplex: