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

Conference utrop1::linkworks_v3

Title:LinkWorks V3.0 Notes Conference
Notice:LNX_APO = APO issues, LINKWORKS_V3 = V3.0 issues
Moderator:tacklr.apd.dec.com::TACK_Lm::TACK_L
Created:Tue Jun 28 1994
Last Modified:Fri Jun 06 1997
Last Successful Update:Fri Jun 06 1997
Number of topics:2269
Total number of notes:8338

2143.0. "Working LNX failover, high availability solutions" by TAV02::YAKIR (Yakir Lavie, ISRAEL DTN 882-3327) Tue Feb 25 1997 14:05

    Hi all,
    
    I thought it may be useful for the LNX community at large to group
    together in one place the accumulated experience with actual, working
    solutions which allow for LNX high availability, failover etc.
    
    I know of some such efforts going on around the globe, so would be
    thankful if others joined in to describe what their customers are doing
    about this painful (?) subject.
    
    thanks,
    
    Yakir
T.RTitleUserPersonal
Name		DateLines
2143.1In IsraelTAV02::YAKIRYakir Lavie, ISRAEL DTN 882-3327Tue Feb 25 1997 14:1112
    Here in Israel we have 1 customer who has implemented the DECsafe
    solution (made by Michael Balomiri @AUI, see Linkworks note # 2105 for
    example). This is working in a dual Alphaserver 1000 using Digital
    Unix, Oracle, LNX & DECsafe. This is working in production for some 4
    months now. We have had 2 failures with this system, in which we found
    out the shortcomings of this solution.
    
    Another customer we have has implemented a "cold" failover on his
    RS6000 servers, running AIX, Oracle and HACMP. By "cold" I mean that
    when the primary LNX server fails, another server assumes its identity,
    disks etc. to resume operation. This solution will go into production soon.
2143.2in switzerlandGENIE::16.184.48.153::genie::tschanzTue Feb 25 1997 17:5154
Here in Switzerland we have three customer who has implemented the 
DECsafe solution. We developed your own solution with the base 
Documentation from Michael Balomiri.

We are also implementing now DIGITAL NSR with DECsafe for the LNX 
environement.

Next Release: Oracle Paralell-Server with DIGITAL Memmory-Channel
              and a failover with the LNX WEBConnector this year.

1. Customer small environment for 80 users:
============================================
DIGITAL Unix V4.0
Oracle V7.3.3
1 Alpha 2100 Server
1 Alpha 3800 Server
1 LNX Cell V3.08
1 HSZ-40
Productive since 2 months with DIGITAL Unix V4.0
Productive since 8 months with DIGITAL Unix V.3.2

2. Customer for 900 users:
==========================
DIGITAL Unix V3.2
Oracle V 7.1.6
2 Alpha 2100 Server in Rackmount
2 LNX CEll with two Communication Cell V3.08
2 HSZ40
Productive since 10 months

3. Customer for 1200 users on DECsafe 600 on dedicated Server in 
planning:
================================================================
DIGITAL Unix V3.2
Oracle V 7.2.3
2 Alpha 1000 Server
4 LNX CEll with one Communication Cell V3.08
2 HSZ40
Productive since 6 months


open Problems on Server:
========================
- Hostnames from cells are hardcoded in the DB 
- no alias support for instmgmt, instcell
- you cannot stop and starting only one cell if you have two cells
  on a Server.

open Problems on Client:
========================
environ.ini take the real hostname, not the alias.
hack on 16bit client with readonly environ.ini.

Toni