| T.R | Title | User | Personal
Name | Date | Lines |
|---|
| 1841.1 | | KITCHE::schott | Eric R. Schott USG Product Management | Fri Jan 24 1997 09:37 | 10 |
| Hi
See http://www-unix.zk3.dec.com/www/clusters.html#disaster
It will tell you to contact Steve Horn. It will also point to
some other information...
regards
Eric
|
| 1841.2 | CSS FibreChannel solution | 16.181.128.8::jpt | FIS and Chips | Mon Jan 27 1997 06:01 | 16 |
|
Last week we wrote three tenders for different customers. Solutions
are FibreChannel based CSS tailored Available Server clusters.
It takes 2-3 working days from CSS to propose a solution based on
detailed specification for customer needs. In Europe Annecy CSS
helps you to quote FC based clusters. They have been very
responsive to our needs, and it's about time, as we have lost
several cases lately not having anything to offer in FC world.
CSS solution is based on Seagate FC disks, Genroco Controllers and
Drivers, Gadzoox Hub and MIA. It gives you full SCSI address mapping
(using LUN's aka HSZ family), it is full SCSI-2 compliant, supports
target mode and emulates KZPSA functionality.
-jari
|
| 1841.3 | CSS to the rescue! | TROOA::MSCHNEIDER | [email protected] | Sat Feb 08 1997 22:13 | 2 |
| Yes from the powerpoint presentation available on the web page it looks
like CSS has put together a very nice solutions.
|
| 1841.4 | | KITCHE::schott | Eric R. Schott USG Product Management | Sun Feb 09 1997 21:36 | 6 |
| Hi
I suggest you speak with Steve Horn. The CSS solution is not
qualified with ASE. It hopefully will be done in the future, but
you should discuss you customer needs with Steve Horn asap.
|
| 1841.5 | Some caveats.... | NETRIX::"[email protected]" | Steve Horn | Wed Feb 12 1997 11:53 | 16 |
|
We are indeed planning on testing the CSS Annecy Fibrechannel configuration
with ASE. Firm dates are not yet available, but I will post them here when
they are (as well as on the DT Web page).
But here is a significant caveat: if we test this configuration with ASE and
it passes totally...you still do not have a true Disaster Tolerant solution.
You do have a Highly Available cluster that can be stretched over a longer
distance...but without a 'DT Safe' mirroring method (see the web page for
definition) you have some significant exposures. Indeed, if you use the
database replication utilities (which are DT Safe) for updating the backup
site
you don't need the Fibre!!
Steve
[Posted by WWW Notes gateway]
|
| 1841.6 | Sizing the problem | ANNECY::CHABORD_D | Dominique CHABORD @AEO | Thu Feb 13 1997 10:16 | 34 |
| Hi,
RE .0: You can get first info on what is delivered on projects in
aeobn0::eic$tn:[ha_osf.disastol]
as well as a documentation of known limitations.
RE .-1:
I am confirming what Steve says: With LSM, there is potential loss
of data in case of disaster. For those interested in knowing how
much and how often, please keep on reading.
If you imagine a disaster like a bomb, there is no problem, because
the computer stops at the same time as the disks.
If you imagine a disaster starts as a sequence of failures, you
have several cases:
- the computer fails first: no problem
- local disks fail before link to remote disks: no problem
- link to remote disks breaks first: here is the problem,
because the computer will write data on the local disk only.
In this third case, there is a gap between the failure and the
"engine stop". I imagine a period of a few seconds, but it is
enough to say that it is not DT safe.
All our customers so far have accepted this risk. For those who
wouldn't, one can complement what we've done or study an
alternative, like data replication at db level. If we make progress
we'll post it.
dominique
|