[Search for users] [Overall Top Noters] [List of all Conferences] [Download this site]

Conference 7.286::fddi

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

83.0. "Concurrent update on 2 remote locations ?" by KETJE::ROBBENS (Big animals move slow...) Tue Jun 26 1990 11:24

    
    I was sent to this notesfile via the Cluster-notesfile.
    Included is the reply pointing to this conference �nd the question.
    We are waiting for hints, solutions, proposals, ideas etc.
    
    Thanks,
    
    Luc.
    
    
================================================================================
Note 1644.1         Concurrent update on 2 remote locations ?             1 of 1
VCSESU::BOWKER "Joe Bowker, KB1GP"                    6 lines  26-JUN-1990 08:00
                             -< VAXclusters/FDDI >-
--------------------------------------------------------------------------------
    See the notes conference on MOUSE::FDDI. I forget where the note
    is but  look for the one on VAXclusters/FDDI. This might meet your
    customer's needs. (VAXclusters/FDDI will be program announced at
    DECworld.)
    
    Joe

            <<< ELKTRA::NOTESPUBLIC:[NOTES$LIBRARY]CLUSTER.NOTE;1 >>>
                              -< + VAXclusters + >-
================================================================================
Note 1644.0         Concurrent update on 2 remote locations ?            1 reply
KETJE::ROBBENS "Big animals move slow..."            28 lines  26-JUN-1990 03:30
--------------------------------------------------------------------------------
    Hello Cluster-experts,
    
    One of our prospects wants us to give him a solution for the following
    problem:

    He is going to start his application with a cluster running Rdb
    (anyway, let's hope so as it's still a prospect...). Within two years,
    for reason of security, the two CPU's will be separated over two
    different locations in town, at a distance of about 4 miles, each CPU
    having its own set of discs, each having his own Rdb database. 
    The problem is that those two Rdb-databases will have to be an exact
    copy of each other, while the two CPU's will be updating the database
    independently. In case of breakdown of one of the CPU's, the other one
    will execute all the updates, until number one comes up again. From then
    on, the databases will have to reflect the same situation again, each
    one acting as updater of the data again.
    The prospect refuses a Master-Slave model where the slave simply
    re-executes the update on his own database: each CPU has to be a
    'master' and each one has to be a 'slave' at the same time.

    Any suggestions hard and/or soft ? Even hints about products still 
    under development are welcome, as for the first to years, it will be a
    real cluster.

    Regards,

    Luc. 
T.RTitleUserPersonal
Name
DateLines
83.1HYEND::BLYONSVAXcluster SASETue Jul 03 1990 11:4410
	- VMS 5.4 will provide Host-based shadowing for cluster storage

	- FDDI will provide a long(er) distance cluster interconnect

	I don't know about 4 miles but concurrent but separated shadowed
	data is going to available in the not too distant future.

	Does that solve your problem?

		Bob L.
83.2is this what you meant?ZPOV03::HWCHOYFE110000Wed Jul 04 1990 01:2121
    re .-1
    
    is this what you meant?
    
    					where x's are members of a shadowset
    
              ___hsc--o-o-x-o disk farm              ___hsc--o-x-o-o
             /                                      /
        SC  *-----VAX...                           *-----VAX...
         	   |                                      |
            	   |                                      |
            	   |                                      |
    =====================================================================
				FDDI
    
    
    Do you mean the 2 sets of CI-cluster can be in a single VAXcluster? And
    that shadowing is supported across the 2?
    
    Rgds,
    hw
83.3Yeh, that's what he means, but don't make any promises you can't keepCVG::PETTENGILLmulpThu Jul 05 1990 03:2418
Yes, that's what he means.

But, there are a number of questions that remain unanswered:

Will the timing of VMS support for FDDI, FDDI adapters, and performance
optimizations and a whole bunch of questions about delays and reliability
be resolved in time for this customer to upgrade to the required version of
VMS in that time frame?  Two years isn't a very long time when you consider
the total length of the product design, implementation, test, release, customer
integration, test, and deployment.

Will public carrier service be available and will DEC have the components in
place to support the distance.  If the customer is in NYC, then in two years
he may be able to get fiber service and they may be fiber `modems' and
other stuff to connect two FDDI segments over that distance.  This application
is probably not a candidate for TransLAN bridges and T1!

I wouldn't make any promises.
83.4HYEND::BLYONSVAXcluster SASEThu Jul 05 1990 09:548
	The answer is even better than the diagram in 83.2, you don't
	necessarily need the Star couplers and HSC storage controllers
	at one or both locations.

	But, as .3 said, it takes time to implement all these planned
	features and two years to get it all on-line might be tight.

		Bob L.
83.5now I'm getting confused...ZPOV03::HWCHOYFE110000Sun Jul 08 1990 13:0321
    re .last few,
    
    thanx for the warning...
    
    this is a military customer, so they'll be responsible for getting the
    fiber laid (they probably wouldn't tell us where the physical cables
    run). They need backup sites that is >30km apart, with about 200
    workstations generating db transactions to the central store. Our
    thinking is that, if volumes can indeed be shadowed across that
    distance on FDDI, than the entire business would be ours (we're talking
    about 200+ DECstation 5000s and 110 TPS of capacity at the central
    store and backup site).
    
    I desperately need to know what the plans are for this space. Can
    anyone provide a name that I can contact to source for more info? I
    will get my management to get in touch if it's way above my head.
    
    thanx,
    hw
    
    Singapore SWS
83.6product management for FDDI clustersHYEND::BLYONSVAXcluster SASEMon Jul 09 1990 10:468
	RE: .5

    Start with Fred Howell on node LESCOM::  or Gailyn Casaday on node HYEND::

    They are the two product management managers involved with FDDI clusters.

	Regards,
		Bob L.
83.7ZPOV03::HWCHOYFE110000Tue Jul 10 1990 00:125
    re .-1
    
    Thanx for the names.
    
    hw