SUMMARY: High Availavility Products

From: mjbueno@nec.com.ar
Date: Tue Apr 21 1998 - 15:34:58 CDT


First of all thaks to:

david@bae.uga.edu
DenveSC@europe.sortek.com
KWhite@talisman-energy.com
Rav@otk.otk.intec.ru
shankar.kanabiran@ac.com
reineman1@raiders.llnl.gov
jason.axley@attws.com
SROTHENB@montefiore.org
sai@pagemart.com
Colin_Melville@mastercard.com

for your prompt replies.

 --------------------------------------------------------------------------
 --------------------------------------------
Original Question:
Hi Sun Managers,

I need information about High Availavility Software Products for Solaris
2.5.1/2.6.
I'll summarize.

Thanks in advance !!

 --------------------------------------------------------------------------
 --------------------------------------------
Suggested Products:

 - QualixHA+ http://www.qualix.com/
 - Veritas FirstWatch http://www.veritas.com/
 - Sun's HA http://www.sun.com/clusters/

I received different opinions regarding which product is better. But the
general opinion
is that Qualix and Veritas are very similar and that Sun's HA software is
pretty young.

 --------------------------------------------------------------------------
 --------------------------------------------

Seth (SROTHENB@montefiore.org) also send me useful information about
important issues to be considered. He wrote:

Some things you should look for:
1. No kernel modification. I don't know what
the benefits of modifying the kernel are, but I
am more comfortable NOT modifying it. Solaris
provides many tools to support HA without mods.

2. Asymetric configuration. The two hosts can be
completely different. You can fail over from
an SS1000E with 4 CPUs to a Sparc20 with 2 CPUs.
All you have to do is configure it.
(That's the hard part of course:-(

3. A vendor who talks about journalling or logging
the file systems. Qualix refused the answer the
question whether we need logging (SDS) when we
put together our system. When I took a class on
SDS, the instructor said there is no question.
Without logging, a failover corrupts the disk,
and fsck can take 2 hours. HA does you no good.

4. Support. It was only after several managed
failovers that I figured out why we end up
having to crash the primary to trigger a failover.
Better documentation or support might have helped.
Cron jobs running on shared disk must be stopped.
 --------------------------------------------------------------------------
 --------------------------------------------

THANKS TO EVERYONE WHO RESPONDED !! :-)
Best Regards,

        Maria

 --------------------------------------------------------------------------
 ------------------------------------
Maria Jose Bueno

Hardware and Software Development Department
Technology Division
NEC Argentina S.A

Av. San Martin 5020 (1602) Florida
Bs. As - Argentina
Phone: (54 - 1) 730-6062
Fax: (54-1) 730-6060
e-mail: mjbueno@nec.com.ar
 --------------------------------------------------------------------------
 ------------------------------------



This archive was generated by hypermail 2.1.2 : Fri Sep 28 2001 - 23:12:38 CDT