domenica 27 gennaio 2008 19:24i have task from my comp, about... "how to clustering Oracle 9i ? can you use the CCS?"
as far as i know that, oracle can be a cluster using RAC for Oracle 9i that found with Release 2.... well thatssss d challenge
now... i've got my test lab running for CCS with
4 Node (Win CCS 2003 x64) +
1 Headnode (Win CCS 2003 x64) +
1 ServiceNode = ActiveDirectoryDC/DHCP/DNS (Win 2003 Entrpc x32)
all works fine.... for 2 day workaround )
for simply thing i've made my web server running on ServiceNode
then i create for job on headnode for apache and mysql....
i just use the service from servicenode of apache then edit few things on http.conf
... and it works i've got every node work on it every on request for web server..
i can see it on CC Administrator -> System Monitor
now when the same method i use to Oracle Services (for d record Oracle 9i Release 2 installed on ServiceNode)...
not working at all..
after i googling for a while.. looking for Ora services, i found that "The OracleOraHome92TNSListener service is only needed if the database will be accessed remotely" --> http://articles.techrepublic.com.com/5100-22-5224417.html
so.. i create job and task... set command line with \\SERVICENODE\Oracle\ora92\bin\...exe (i forget bout filename)
and.. it running, but when i try to select data for about 1000000 line... it doesn't look that every node work on it like on web server they did.
now... is there someone out there.. have the same problem ???
need help here ?
Tutte le risposte
venerdì 22 febbraio 2008 20:08Moderatore
Hi rez, there are several types of clustering that people talk about. Disk clustering for storage, geo-clustering so people around the world have fast access to the same resources or for disaster recovery, load balancing clustering for application availability and performance, and computational clustering for high performance calculations.
What you are trying to do is the 'load balancing' type of clustering. That is, you want your database and web server running parallel copies on multiple machines with incoming requests 'balanced' across the different machines to keep responsiveness up.
The Compute Cluster Pack, and the upcoming version 2 HPC Pack, are for setting up computational clusters. These are clusters where the individual computers (compute nodes) work together to solve a complex problem (aerodynamics of a car body) or a bunch of closely-related problems (doing a simple math function with millions of different inputs).
So, unfortuantely, you've gone down the wrong path a little bit. What you should do is check out the load balance and failover clustering features within the operating system itself, as well as the documentation for Oracle and Apache. Apache is easy to set up in a clustered way, I'm not sure how to do Oracle though.
- Contrassegnato come risposta Don PatteeModerator mercoledì 25 marzo 2009 23:29