[Date Prev][Date Next]
FW: New to list: question about High Availability
- To: "Open LDAP mailing list (E-mail)" <OpenLDAP-software@OpenLDAP.org>
- Subject: FW: New to list: question about High Availability
- From: "Curtis Bond" <Curtis@wagerworks.com>
- Date: Wed, 18 Jul 2001 09:34:12 -0700
- Content-class: urn:content-classes:message
- Thread-index: AcEO/zPj0Wjn5at0SuaExhEzA6M++AAqD8jQ
- Thread-topic: New to list: question about High Availability
From: Curtis Bond
Sent: Tuesday, July 17, 2001 1:43 PM
To: Open LDAP mailing list (E-mail)
Subject: New to list: question about High Availability
After searching the archive I was hoping to find out what people are
currently doing for High Availability for OpenLDAP.
Our needs are:
Large number of transactions per second.
No downtime (or seamless fail over) for BOTH reads and writes
No need to slave but need redundant servers with identical databases.
Basically we are a high volume e-commerce site and are using LDAP for
transaction authentication and user login. We have a ratio of about 90
reads to every write with current volume of 300,000+ transactions per
day which will increase to 1,000,000+ once the site is ramped up.
I've been searching the HA sites for answers but so far nothing without
switching LDAP base. Sure there is Veritas, Netscape, etc but I'm
looking to find a solution in the open-source community.
The big problem is the replication. As the number of changes happening
would seem to slow down the system if the data was being replicated at a
rate that would be sufficient that if a server went down you'd maybe
only lose the data for a few transactions.
Sharing the database between multiple servers seems to be one answer but
I'm not sure anyone has done this. As long as only one server has write
permissions it shouldn't be an issue.
Any thoughts or links would be greatly appreciated.
Sr. Systems Administrator
2339 Third St. Fourth Floor
San Francisco, CA 94107