Re: can run HDR and ER simultaneously on the same server IDS 9.20??
It is a very good idea to use HACMP for high availability purposes (all servers in the cluster must be IBM's machines with AIX), but even if you use HACMP you must have a database replication layer/estrategy (HDR or ER), because HACMP is at OS level and it does not replicate data in databases, remember that if chunks are raw devices, the OS does not know the format of the data, and even if chunks are cooked files, it is not a good idea to replicate them at os level, a few days ago some members of the group discussed about replicating via RSYNC, check out the list archive . High Availability Cluster MultiProcessing is a very good product for disaster recovery with a very small time of downtime (minutes or even seconds), but it needs the data to be ready to be used in the secondary server, and you have to ensure it at database (or application) level. I hope this help you Regards ----- Original Message ----- From: "Data Goob" <datagoob@hotmail.com> To: <informix-list@iiug.org> Sent: Wednesday, July 28, 2004 6:14 PM Subject: Re: can run HDR and ER simultaneously on the same server IDS 9.20?? > Check out the IBM product : HACMP/XD > > Look for the white paper "Automated Recovery Management > With IBM HACMP/XD and PPRC" on the www.ibm.com website. > > The "XD" is significant, meaning eXtended Distance. > > It is not yet available for Linux last I heard but is > available on an AIX platform. If you go AIX on Power5 > they're running around $25K USD for a p550 machine. > ( HACMP/XD costs additional. ( YMMV ) ) > > Cristian Zaharioiu wrote: > > Hello all, > > > > I administrate about 80 servers running IDS 9.20 and SCO Unixware > > 7.1. At this moment we use ER: one server is root server and the other > > are leaf servers. > > We build a disaster recovery center and we want to replicate whole > > databases on the root server on this location. how can we do this? Can > > using HDR for this? Is their > > likely to be conflict between the ER and HDR threads on the same > > server(machine)? > > The distance between main location and disaster recovery location is > > about 200 km.It's possible HDR at this distance? > > > > Thank you very much! > > > > Regards, > > Cristian > > sending to informix-list
Another solution would be to migrate databases to a mainframe (we have thought at IBM ZSeries) and to use a storage IBM (storage are own OS for replication) at main location and also another mainframe with storage at disaster recovery center. In this case we have thought to use the storage OS for replication of databases. It's a good idea? Could we have problems? It's possible that on secondary mainframe Informix not to know the format of the data if we use this kind of replication at storage OS level? The chunks are cooked files. "Francisco Roldan" <f_roldan@admcentralagricola.com.gt> wrote in message news:<ce9o66$7lk$1@news.xmission.com>... > It is a very good idea to use HACMP for high availability > purposes (all servers in the cluster must be IBM's machines with AIX), > but even if you use HACMP you must have > a database replication layer/estrategy (HDR or ER), > because HACMP is at OS level and it does not replicate data in > databases, remember that if chunks are raw devices, > the OS does not know the format of the data, and even > if chunks are cooked files, it is not a good idea to replicate them > at os level, a few days ago some members of the group > discussed about replicating via RSYNC, check out the list archive . > > High Availability Cluster MultiProcessing > is a very good product for disaster recovery with > a very small time of downtime (minutes or even seconds), > but it needs the data to be ready to be used in the secondary server, > and you have to ensure it at database (or application) level. > > I hope this help you > > Regards >
If you can migrate the databases to a mainframe, then you could also upgrade the database version to 9.4. Generally, I would want to take the easiest/most familiar solution. There is a significant mind-set difference between a main frame environment and a UNIX environment. "Cristian Zaharioiu" <cristizaharioiu@hotmail.com> wrote in message news:929f826d.0407290541.4557d692@posting.google.com... > Another solution would be to migrate databases to a mainframe (we have > thought at IBM ZSeries) and to use a storage IBM (storage are own OS > for replication) at main location and also another mainframe with > storage at disaster recovery center. > In this case we have thought to use the storage OS for replication of > databases. > > It's a good idea? > Could we have problems? It's possible that on secondary mainframe > Informix not to know the format of the data if we use this kind of > replication at storage OS level? > The chunks are cooked files. > > > "Francisco Roldan" <f_roldan@admcentralagricola.com.gt> wrote in message news:<ce9o66$7lk$1@news.xmission.com>... > > It is a very good idea to use HACMP for high availability > > purposes (all servers in the cluster must be IBM's machines with AIX), > > but even if you use HACMP you must have > > a database replication layer/estrategy (HDR or ER), > > because HACMP is at OS level and it does not replicate data in > > databases, remember that if chunks are raw devices, > > the OS does not know the format of the data, and even > > if chunks are cooked files, it is not a good idea to replicate them > > at os level, a few days ago some members of the group > > discussed about replicating via RSYNC, check out the list archive . > > > > High Availability Cluster MultiProcessing > > is a very good product for disaster recovery with > > a very small time of downtime (minutes or even seconds), > > but it needs the data to be ready to be used in the secondary server, > > and you have to ensure it at database (or application) level. > > > > I hope this help you > > > > Regards > >
Madison Pruet wrote: > If you can migrate the databases to a mainframe, then you could also upgrade > the database version to 9.4. Generally, I would want to take the > easiest/most familiar solution. There is a significant mind-set difference > between a main frame environment and a UNIX environment. Or you can run IDS on Linux on the Mainframe... Just to add a bit of confusion on a friday afternoon :) Regards.
Not on 9.2 "Fernando Nunes" <spam@domus.online.pt> wrote in message news:2msq6kFqjk3gU1@uni-berlin.de... > Madison Pruet wrote: > > If you can migrate the databases to a mainframe, then you could also upgrade > > the database version to 9.4. Generally, I would want to take the > > easiest/most familiar solution. There is a significant mind-set difference > > between a main frame environment and a UNIX environment. > > > Or you can run IDS on Linux on the Mainframe... > Just to add a bit of confusion on a friday afternoon :) > > Regards.