| Title: | FDDI - The Next Generation |
| Moderator: | NETCAD::STEFANI |
| Created: | Thu Apr 27 1989 |
| Last Modified: | Thu Jun 05 1997 |
| Last Successful Update: | Fri Jun 06 1997 |
| Number of topics: | 2259 |
| Total number of notes: | 8590 |
A customer is planning to build a backup data center. They presently
have in one datacenter with a 7620, a 6610 clustered via a star coupler,
and a 4400 hanging off the 7620 via DSSI (hence a mixed VAXcluster).
They have in place an FDDI ring between the current datacenter and the
proposed backup site. Both the 6610 and the 7620 are attached to the
FDDI ring via FDDI adapters through FDDI concentrators.
ALL three systems are on Ethernet (required for mixed cluster). This
Ethernet is also bridged to the FDDI ring. They have a quorum disk
available on the HSC.
Here is my question(s):
They intend to *break* the CI link from the 6610 to the star coupler to
simulate the 6610 residing at the backup site. Will this even work? can
a cluster be configure such? MDF talks of multi-lobe clusters where
two clusters are joined over FDDI. Here the customer is proposing
placing two node at one site and the other at a different site. If this
is possible can I get away just one system disk? What about quorum?
Cheers,
Peter Ball
| T.R | Title | User | Personal Name | Date | Lines |
|---|---|---|---|---|---|
| 1140.1 | SSDEVO::PARRIS | The SLED is dead, long live RAID | Fri Nov 05 1993 14:59 | 29 | |
> They intend to *break* the CI link from the 6610 to the star coupler to > simulate the 6610 residing at the backup site. Will this even work? can > a cluster be configure such? Yes, this can work. The OpenVMS MSCP Server needs to be loaded on at least one of the local systems for the remote system to be able to access the HSC disks over FDDI (and the server loaded on the other end also if remote disks are to be accessed by the local systems). > Here the customer is proposing > placing two node at one site and the other at a different site. > Can I get away with just one system disk? The 6610 at the remote site will need its own local system disk (since it can't boot as a satellite). Cluster-common files can be placed on another disk (or shadow set), so having to have two systems disks isn't too bad in practice. > What about quorum? Having a quorum disk on an HSC at one end won't help much if that end fails. I'd say give one vote to each of the three systems, and don't have a quorum disk. This makes the site with two systems the primary site; the other site can't automatically continue operation if the link between the sites fails. See also ISEQ::MDF. | |||||