Re: Cluster with informix.

From: "Centro Medico Paitilla" <cmp@pty.com> > >We are evaluating an HA (Cluster) solution from SUN >Microsystems and IBM, but i have some questions about the >capacity of Informix server of work in that environment. I hope that >someone can help me with this. Informix doesn't really support Sun Clusters, or is it that Sun Clusters don't support Informix? Depends on who you talk to. But I've just been to a similar debate recently, and the upshot in our case was that we are going to Enterprise Replication because we need to propogate our database from a primary server to a backup server and to a reporting box and to a development box and.... The key with ER is to get your database designed correctly from the get-go for ER. If you're in that happy position, I think ER is a great way to solve this. Another possible solution, which we currently use, is to use HDR for fault tolerance (and it is very tolerant of our faults! :) and do some of the reading of data from the secondary. So we get load balancing on that basis. But that might be difficult if you're not using ODBC. >1. Do I have to buy Informix Paralell Server or IDS.2000? It is an >OLTP environment... There is no such thing as Informix Parallel Server, that sounds like an Oracle-ism. You might be thinking of Informix XPS or EDS, and that isn't for OLTP, it's for decision support. So you need IDS.2000. >2. With the case that we have two machines in cluster using a >cluster software from a machine provider, Can Informix servers work >together sharing 50% of the work? Yes and no. You can't run in a Sun cluster, but you can use Enterprise Replication to allow both machines to do half the work each. >3. If I have an application running on 4GL (RDS), Do I have to buy >licences of informix products for both machines? Yes. >4. Is Informix data replication capability more efficient that HA >environment based on cluster software from a machine provider? I guess so, because it's supported. :-) Another possible solution, which may be better overall, is to have a three-box solution. If you were looking at two (say) E4500's, you might replace that with 3 E450's, where one would be just for 4GL applications, one would be a primary server and one a secondary, read-only server. You could do all your data entry onto the primary, run your reports against the secondary and have all the 4GL load on the third box. YAMMV. How much load are you expecting to put on the server? Clustering always seems a bit of an extreme solution... If you're looking at The Other Database as a potential clustering solution, let me know. I might be able to dredge up some info on the Distributed Lock Mangler that should help you to change your mind. :-) ________________________________________________________________________ Get Your Private, Free E-mail from MSN Hotmail at http://www.hotmail.com