nagy%warner.hepnet@LBL.ARPA (07/12/87)
>A shared disk drive does not a Vaxcluster make. The only reason people >dual-port disks on VMS, I think, is so that they can be manually switched >(without having to move cables) over to machine B if machine A goes down. Actually, static dual-porting works quite nicely as a failover system with HSCs or VAXes with UDA50s/KDB50s (BI controllers) in Big VAXClusters. I even tried it on a live system once we got everything hooked up and the correct software versions installed (i.e. we finally moved from V3.7 to V4.2) and was astonished at how smoothly it worked. We have yet to test, but plan to, a similar failover system between MicroVAXes in an LAVC. It should work since it just involves the VMS software (and setting an allocation class for the nodes connected to the disks) and an LAVC runs full VMS (i.e., our RA60 system disk would boot an 8650!). >More seriously, Dec says you can't plug an RA81 into a microVax (at least >our uVax's) at all - seems Dec screwed up such that the necessary cabling >would break FCC rules. I assume that that means there would be "too much" >RF leaking out of the cabinet. Wrongo. All our MicroVAXes are BA23 systems which have been rack mounted (several were bought as pedestal systems, some even as VAXStation-II/RCs) with BA23-CC expander boxes. The expander boxes are needed for the KDA50 controllers (QBus controllers for DSA disks) since the KDA50 uses LOTS and LOTS of power and the power supply in the BA23 can't run it and the CPU and the memory (16 MB as dual 8 meg boards), TK50 controller, DEQNA and whatever else is stuffed into that BA23. I believe that KDA50s can be added to BA123 (World Box) systems with appropriate distribution panel kit to prevent that RF leaking. When the KDA50 was first announced, it was NOT available for the BA123s. = Frank J. Nagy = Fermilab Research Division EED/Controls = HEPNET: WARNER::NAGY (43198::NAGY) or FNAL::NAGY (43009::NAGY) = BitNet: NAGY@FNAL = USnail: Fermilab POB 500 MS/220 Batavia, IL 60510