 | |
Log shipping with clustering
microsoft.public.sqlserver.clustering microsoft.public.sqlserver.clustering
 |
| | Thread Tools | Display Modes |
#11
 
| |||
| |||
|
Re: Log shipping with clustering -
09-10-2004
, 01:32 PM
|
Currently there is no automagic client redirection. You can handle this in a variety of ways. You can put the server portion of the conneciotn string in a registry entry or an XML document and push changes otu to all the clients. You can do a DNS alias redirect, but as you noted, there is the client caching issue. This is one reason clustering is so useful. The clients reconnect using the same connection string after a failover. -- Geoff N. Hiten Microsoft SQL Server MVP Senior Database Administrator Careerbuilder.com I support the Professional Association for SQL Server www.sqlpass.org "barbara" <barbara_2003 (AT) hotmail (DOT) com> wrote in message news:5cd01256.0409100354.23e1b6a4 (AT) posting (DOT) google.com... Thanks for the good explanation. How do we change the client configuration to point clients from SQLCLUSTER to node1? Oracle uses LDAP/ONAMES for name resolution which is easier to change. Do we have any for sql server? Is DNS name change is only option available? How does it take care of caching at client level? "Luca Bianchi" <rightjoinREMOVE_THIS (AT) hotmail (DOT) com> wrote in message news:<2qd1hrFueik4U1 (AT) uni-berlin (DOT) de>... barbara wrote: Luca, Thanks for the info. I've looked at all these articles before. If I create standby on same server as cluser node (business doesn't want to spend on another server), how do I failover to this? How does the client connects to standby? What name does it uses because the cluster node also exists on the same server. I agree completely with Geoff and Greg about their consideration. High Availability has a cost. If you don't consider it, probably you don't need a cluster solution. By the way you have to refer to the clustered instance with its VIRTUAL SERVER NAME that is different by the name of two node machine. The stand alone instance could be referred by the traditional host name. Imagine that you have a clustered node called NODE1 and the other node called NODE2 (either with its own IP address). When you setup a clustered (at operating system level) you have to assign, for the clustered, one VIRTUAL SERVER NAME and one VIRTUAL IP ADDRESS (of the same class of both node). When you install the virtual server instance of SQL Server you have to provide another VIRTUAL SERVER NAME and another VIRTUAL IP ADDRESS for SQL Server. For example The machine that form the cluster are: NODE1 (10.0.0.1) NODE2 (10.0.0.2) When you create a clustered at Windows 2000/2003 level you can assign the following name and ip address to the virtual server W2KCLUSTER (10.0.0.3) And then you setup SQL Server in a clustered instance creating a new virtual server name with its virtual ip address SQLCLUSTER (10.0.0.4) If you install a stand-alone instance on NODE1 you can refer to the clustered instance with the clustered instance name (that could be run on NODE1 or NODE2) using its name or ip address (SQLCLUSTER or 10.0.0.4). If you would connect to the stand alone instance you can connect to it without considering the virtual server name but using NODE1. Bye |
#12
| |||
| |||
|
|
Does microsoft know this problem? In enterprise environment, it is very hard to push the DNS change to all clients. When they offer log shipping, there should be an easy way for client redirection. I feel this is is the biggest drawback with sql server. How do you push XML or registry to all clients? Do we need to have list of all clinet machine IPs? Any other ideas in this subject is well appreciated. Thanks "Geoff N. Hiten" <SRDBA (AT) Careerbuilder (DOT) com> wrote Currently there is no automagic client redirection. You can handle this in a variety of ways. You can put the server portion of the conneciotn string in a registry entry or an XML document and push changes otu to all the clients. You can do a DNS alias redirect, but as you noted, there is the client caching issue. This is one reason clustering is so useful. The clients reconnect using the same connection string after a failover. -- Geoff N. Hiten Microsoft SQL Server MVP Senior Database Administrator Careerbuilder.com I support the Professional Association for SQL Server www.sqlpass.org "barbara" <barbara_2003 (AT) hotmail (DOT) com> wrote in message news:5cd01256.0409100354.23e1b6a4 (AT) posting (DOT) google.com... Thanks for the good explanation. How do we change the client configuration to point clients from SQLCLUSTER to node1? Oracle uses LDAP/ONAMES for name resolution which is easier to change. Do we have any for sql server? Is DNS name change is only option available? How does it take care of caching at client level? "Luca Bianchi" <rightjoinREMOVE_THIS (AT) hotmail (DOT) com> wrote in message news:<2qd1hrFueik4U1 (AT) uni-berlin (DOT) de>... barbara wrote: Luca, Thanks for the info. I've looked at all these articles before. If I create standby on same server as cluser node (business doesn't want to spend on another server), how do I failover to this? How does the client connects to standby? What name does it uses because the cluster node also exists on the same server. I agree completely with Geoff and Greg about their consideration. High Availability has a cost. If you don't consider it, probably you don't need a cluster solution. By the way you have to refer to the clustered instance with its VIRTUAL SERVER NAME that is different by the name of two node machine. The stand alone instance could be referred by the traditional host name. Imagine that you have a clustered node called NODE1 and the other node called NODE2 (either with its own IP address). When you setup a clustered (at operating system level) you have to assign, for the clustered, one VIRTUAL SERVER NAME and one VIRTUAL IP ADDRESS (of the same class of both node). When you install the virtual server instance of SQL Server you have to provide another VIRTUAL SERVER NAME and another VIRTUAL IP ADDRESS for SQL Server. For example The machine that form the cluster are: NODE1 (10.0.0.1) NODE2 (10.0.0.2) When you create a clustered at Windows 2000/2003 level you can assign the following name and ip address to the virtual server W2KCLUSTER (10.0.0.3) And then you setup SQL Server in a clustered instance creating a new virtual server name with its virtual ip address SQLCLUSTER (10.0.0.4) If you install a stand-alone instance on NODE1 you can refer to the clustered instance with the clustered instance name (that could be run on NODE1 or NODE2) using its name or ip address (SQLCLUSTER or 10.0.0.4). If you would connect to the stand alone instance you can connect to it without considering the virtual server name but using NODE1. Bye |
#13
| |||
| |||
|
|
Does microsoft know this problem? In enterprise environment, it is very hard to push the DNS change to all clients. When they offer log shipping, there should be an easy way for client redirection. I feel this is is the biggest drawback with sql server. How do you push XML or registry to all clients? Do we need to have list of all clinet machine IPs? |
|
« Previous Thread
|
Next Thread »
| Thread Tools | |
| Display Modes | |
Linear Mode
| |
Powered by vBulletin Version 3.5.3
Copyright ©2000 - 2012, Jelsoft Enterprises Ltd.
Copyright ©2000 - 2012, Jelsoft Enterprises Ltd.