[HECnet] "Mixed" clustering

[Kari Uusim ki]

Yes, indeed, the Qbus is definately the bottleneck here. As also if you do use SCSI disks say through a KFQSA and HSD* controller.
The KZQSA is very limited, IIRC. I'm not aware of any support for SCSI disks (except for CD drives) on it.


Kari



Zane H. Healy wrote:
That's part of what makes the idea interesting.   Of course I suspect you'd
saturate your Q-Bus backplane with it.   IIRC, the backplane can handle
roughly 3MB/s.
I suspect such a solution might be faster than a MicroVAX II with SCSI
drives.   I've not tested the theory as while PCI FDDI cards are cheap
enough, the Q-Bus ones were not the last time I checked.
Zane
On Tue, 29 Sep 2009, Kari Uusim  ki wrote:

About the comparation between FDDI and RD* disks. Theoretically FDDI should be faster, because it is capable of delivering about 12MB/s, compared to the RD* disks which might be able to deliver about 1MB/s.


Kari


Zane H. Healy wrote:


On Tue, 29 Sep 2009, Kari Uusim  ki wrote:

first when you do a disk shadow copy and second when you boot a satellite over the Ethernet.

This is where I've always wanted to play with FDDI.   I'd love to boot a
MicroVAX II off of either a high end VAX or an Alpha via a FDDI link.   I'm
curious as to which would be faster.   Running the MicroVAX II off of native
disks, or over the FDDI link.

Zane