![]() | |
![]() |
| | Thread Tools | Display Modes |
#71
| |||
| |||
|
|
Actually a better definition of the 'shared nothing' model is that although resources (disks, IPs, network names, applications, etc.) can move between nodes in a cluster, one node and ONLY one node has access or "owns" those resources at a time. You can see this with disks in that if you try and access a disk in a cluster owned by another node -- Jeff Hughes, MCSE Senior Support Escalation Engineer Microsoft Enterprise Platforms Support (Server Core/Cluster) "Geoff N. Hiten" <SQLCraftsman (AT) gmail (DOT) com> wrote in message news:OuCnY3ECJHA.1628 (AT) TK2MSFTNGP02 (DOT) phx.gbl... Shared Nothing means that no node or instance is dependent on any resource not on that node or instance. As such, instances cannot access data files owned by another instance. I agree that the terminology "shared disk" is very misleading, as is "Active/Active" and all its variants. Mike Hotek does the rant better than I do on this topic. I always use "multi-connected" disk and emphasize that Clustering arbitrates ownership so only one node actually controls a resource at any given time. -- Geoff N. Hiten Principal SQL Infrastructure Consultant Microsoft SQL Server MVP "Roy Harvey (SQL Server MVP)" <roy_harvey (AT) snet (DOT) net> wrote in message news:5ulab45er6g0o6jh5vuken2466dll25ons (AT) 4ax (DOT) com... On Wed, 27 Aug 2008 09:16:00 +0100, "Edwin vMierlo [MVP]" EdwinvMierlo (AT) discussions (DOT) microsoft.com> wrote: Oh, my understanding was that with active/active that it is load balancing the same database on shared storage and if one fails the other just keeps on trucking. That can't be done? common misunderstanding, Microsoft Failover clustering is a "shared nothnig" cluster model, Microsoft failover clustering is not "shared nothing". The disks are quite clearly shared. The closest SQL Server comes to shared nothing is distributed partitioned views. Roy Harvey Beacon Falls, CT |
#72
| |||
| |||
|
|
Actually a better definition of the 'shared nothing' model is that although resources (disks, IPs, network names, applications, etc.) can move between nodes in a cluster, one node and ONLY one node has access or "owns" those resources at a time. You can see this with disks in that if you try and access a disk in a cluster owned by another node -- Jeff Hughes, MCSE Senior Support Escalation Engineer Microsoft Enterprise Platforms Support (Server Core/Cluster) "Geoff N. Hiten" <SQLCraftsman (AT) gmail (DOT) com> wrote in message news:OuCnY3ECJHA.1628 (AT) TK2MSFTNGP02 (DOT) phx.gbl... Shared Nothing means that no node or instance is dependent on any resource not on that node or instance. As such, instances cannot access data files owned by another instance. I agree that the terminology "shared disk" is very misleading, as is "Active/Active" and all its variants. Mike Hotek does the rant better than I do on this topic. I always use "multi-connected" disk and emphasize that Clustering arbitrates ownership so only one node actually controls a resource at any given time. -- Geoff N. Hiten Principal SQL Infrastructure Consultant Microsoft SQL Server MVP "Roy Harvey (SQL Server MVP)" <roy_harvey (AT) snet (DOT) net> wrote in message news:5ulab45er6g0o6jh5vuken2466dll25ons (AT) 4ax (DOT) com... On Wed, 27 Aug 2008 09:16:00 +0100, "Edwin vMierlo [MVP]" EdwinvMierlo (AT) discussions (DOT) microsoft.com> wrote: Oh, my understanding was that with active/active that it is load balancing the same database on shared storage and if one fails the other just keeps on trucking. That can't be done? common misunderstanding, Microsoft Failover clustering is a "shared nothnig" cluster model, Microsoft failover clustering is not "shared nothing". The disks are quite clearly shared. The closest SQL Server comes to shared nothing is distributed partitioned views. Roy Harvey Beacon Falls, CT |
#73
| |||
| |||
|
|
Actually a better definition of the 'shared nothing' model is that although resources (disks, IPs, network names, applications, etc.) can move between nodes in a cluster, one node and ONLY one node has access or "owns" those resources at a time. You can see this with disks in that if you try and access a disk in a cluster owned by another node -- Jeff Hughes, MCSE Senior Support Escalation Engineer Microsoft Enterprise Platforms Support (Server Core/Cluster) "Geoff N. Hiten" <SQLCraftsman (AT) gmail (DOT) com> wrote in message news:OuCnY3ECJHA.1628 (AT) TK2MSFTNGP02 (DOT) phx.gbl... Shared Nothing means that no node or instance is dependent on any resource not on that node or instance. As such, instances cannot access data files owned by another instance. I agree that the terminology "shared disk" is very misleading, as is "Active/Active" and all its variants. Mike Hotek does the rant better than I do on this topic. I always use "multi-connected" disk and emphasize that Clustering arbitrates ownership so only one node actually controls a resource at any given time. -- Geoff N. Hiten Principal SQL Infrastructure Consultant Microsoft SQL Server MVP "Roy Harvey (SQL Server MVP)" <roy_harvey (AT) snet (DOT) net> wrote in message news:5ulab45er6g0o6jh5vuken2466dll25ons (AT) 4ax (DOT) com... On Wed, 27 Aug 2008 09:16:00 +0100, "Edwin vMierlo [MVP]" EdwinvMierlo (AT) discussions (DOT) microsoft.com> wrote: Oh, my understanding was that with active/active that it is load balancing the same database on shared storage and if one fails the other just keeps on trucking. That can't be done? common misunderstanding, Microsoft Failover clustering is a "shared nothnig" cluster model, Microsoft failover clustering is not "shared nothing". The disks are quite clearly shared. The closest SQL Server comes to shared nothing is distributed partitioned views. Roy Harvey Beacon Falls, CT |
#74
| |||
| |||
|
|
Actually a better definition of the 'shared nothing' model is that although resources (disks, IPs, network names, applications, etc.) can move between nodes in a cluster, one node and ONLY one node has access or "owns" those resources at a time. You can see this with disks in that if you try and access a disk in a cluster owned by another node -- Jeff Hughes, MCSE Senior Support Escalation Engineer Microsoft Enterprise Platforms Support (Server Core/Cluster) "Geoff N. Hiten" <SQLCraftsman (AT) gmail (DOT) com> wrote in message news:OuCnY3ECJHA.1628 (AT) TK2MSFTNGP02 (DOT) phx.gbl... Shared Nothing means that no node or instance is dependent on any resource not on that node or instance. As such, instances cannot access data files owned by another instance. I agree that the terminology "shared disk" is very misleading, as is "Active/Active" and all its variants. Mike Hotek does the rant better than I do on this topic. I always use "multi-connected" disk and emphasize that Clustering arbitrates ownership so only one node actually controls a resource at any given time. -- Geoff N. Hiten Principal SQL Infrastructure Consultant Microsoft SQL Server MVP "Roy Harvey (SQL Server MVP)" <roy_harvey (AT) snet (DOT) net> wrote in message news:5ulab45er6g0o6jh5vuken2466dll25ons (AT) 4ax (DOT) com... On Wed, 27 Aug 2008 09:16:00 +0100, "Edwin vMierlo [MVP]" EdwinvMierlo (AT) discussions (DOT) microsoft.com> wrote: Oh, my understanding was that with active/active that it is load balancing the same database on shared storage and if one fails the other just keeps on trucking. That can't be done? common misunderstanding, Microsoft Failover clustering is a "shared nothnig" cluster model, Microsoft failover clustering is not "shared nothing". The disks are quite clearly shared. The closest SQL Server comes to shared nothing is distributed partitioned views. Roy Harvey Beacon Falls, CT |
#75
| |||
| |||
|
|
Actually a better definition of the 'shared nothing' model is that although resources (disks, IPs, network names, applications, etc.) can move between nodes in a cluster, one node and ONLY one node has access or "owns" those resources at a time. You can see this with disks in that if you try and access a disk in a cluster owned by another node -- Jeff Hughes, MCSE Senior Support Escalation Engineer Microsoft Enterprise Platforms Support (Server Core/Cluster) "Geoff N. Hiten" <SQLCraftsman (AT) gmail (DOT) com> wrote in message news:OuCnY3ECJHA.1628 (AT) TK2MSFTNGP02 (DOT) phx.gbl... Shared Nothing means that no node or instance is dependent on any resource not on that node or instance. As such, instances cannot access data files owned by another instance. I agree that the terminology "shared disk" is very misleading, as is "Active/Active" and all its variants. Mike Hotek does the rant better than I do on this topic. I always use "multi-connected" disk and emphasize that Clustering arbitrates ownership so only one node actually controls a resource at any given time. -- Geoff N. Hiten Principal SQL Infrastructure Consultant Microsoft SQL Server MVP "Roy Harvey (SQL Server MVP)" <roy_harvey (AT) snet (DOT) net> wrote in message news:5ulab45er6g0o6jh5vuken2466dll25ons (AT) 4ax (DOT) com... On Wed, 27 Aug 2008 09:16:00 +0100, "Edwin vMierlo [MVP]" EdwinvMierlo (AT) discussions (DOT) microsoft.com> wrote: Oh, my understanding was that with active/active that it is load balancing the same database on shared storage and if one fails the other just keeps on trucking. That can't be done? common misunderstanding, Microsoft Failover clustering is a "shared nothnig" cluster model, Microsoft failover clustering is not "shared nothing". The disks are quite clearly shared. The closest SQL Server comes to shared nothing is distributed partitioned views. Roy Harvey Beacon Falls, CT |
#76
| |||
| |||
|
|
Actually a better definition of the 'shared nothing' model is that although resources (disks, IPs, network names, applications, etc.) can move between nodes in a cluster, one node and ONLY one node has access or "owns" those resources at a time. You can see this with disks in that if you try and access a disk in a cluster owned by another node -- Jeff Hughes, MCSE Senior Support Escalation Engineer Microsoft Enterprise Platforms Support (Server Core/Cluster) "Geoff N. Hiten" <SQLCraftsman (AT) gmail (DOT) com> wrote in message news:OuCnY3ECJHA.1628 (AT) TK2MSFTNGP02 (DOT) phx.gbl... Shared Nothing means that no node or instance is dependent on any resource not on that node or instance. As such, instances cannot access data files owned by another instance. I agree that the terminology "shared disk" is very misleading, as is "Active/Active" and all its variants. Mike Hotek does the rant better than I do on this topic. I always use "multi-connected" disk and emphasize that Clustering arbitrates ownership so only one node actually controls a resource at any given time. -- Geoff N. Hiten Principal SQL Infrastructure Consultant Microsoft SQL Server MVP "Roy Harvey (SQL Server MVP)" <roy_harvey (AT) snet (DOT) net> wrote in message news:5ulab45er6g0o6jh5vuken2466dll25ons (AT) 4ax (DOT) com... On Wed, 27 Aug 2008 09:16:00 +0100, "Edwin vMierlo [MVP]" EdwinvMierlo (AT) discussions (DOT) microsoft.com> wrote: Oh, my understanding was that with active/active that it is load balancing the same database on shared storage and if one fails the other just keeps on trucking. That can't be done? common misunderstanding, Microsoft Failover clustering is a "shared nothnig" cluster model, Microsoft failover clustering is not "shared nothing". The disks are quite clearly shared. The closest SQL Server comes to shared nothing is distributed partitioned views. Roy Harvey Beacon Falls, CT |
#77
| |||
| |||
|
|
Actually a better definition of the 'shared nothing' model is that although resources (disks, IPs, network names, applications, etc.) can move between nodes in a cluster, one node and ONLY one node has access or "owns" those resources at a time. You can see this with disks in that if you try and access a disk in a cluster owned by another node -- Jeff Hughes, MCSE Senior Support Escalation Engineer Microsoft Enterprise Platforms Support (Server Core/Cluster) "Geoff N. Hiten" <SQLCraftsman (AT) gmail (DOT) com> wrote in message news:OuCnY3ECJHA.1628 (AT) TK2MSFTNGP02 (DOT) phx.gbl... Shared Nothing means that no node or instance is dependent on any resource not on that node or instance. As such, instances cannot access data files owned by another instance. I agree that the terminology "shared disk" is very misleading, as is "Active/Active" and all its variants. Mike Hotek does the rant better than I do on this topic. I always use "multi-connected" disk and emphasize that Clustering arbitrates ownership so only one node actually controls a resource at any given time. -- Geoff N. Hiten Principal SQL Infrastructure Consultant Microsoft SQL Server MVP "Roy Harvey (SQL Server MVP)" <roy_harvey (AT) snet (DOT) net> wrote in message news:5ulab45er6g0o6jh5vuken2466dll25ons (AT) 4ax (DOT) com... On Wed, 27 Aug 2008 09:16:00 +0100, "Edwin vMierlo [MVP]" EdwinvMierlo (AT) discussions (DOT) microsoft.com> wrote: Oh, my understanding was that with active/active that it is load balancing the same database on shared storage and if one fails the other just keeps on trucking. That can't be done? common misunderstanding, Microsoft Failover clustering is a "shared nothnig" cluster model, Microsoft failover clustering is not "shared nothing". The disks are quite clearly shared. The closest SQL Server comes to shared nothing is distributed partitioned views. Roy Harvey Beacon Falls, CT |
#78
| |||
| |||
|
|
Actually a better definition of the 'shared nothing' model is that although resources (disks, IPs, network names, applications, etc.) can move between nodes in a cluster, one node and ONLY one node has access or "owns" those resources at a time. You can see this with disks in that if you try and access a disk in a cluster owned by another node -- Jeff Hughes, MCSE Senior Support Escalation Engineer Microsoft Enterprise Platforms Support (Server Core/Cluster) "Geoff N. Hiten" <SQLCraftsman (AT) gmail (DOT) com> wrote in message news:OuCnY3ECJHA.1628 (AT) TK2MSFTNGP02 (DOT) phx.gbl... Shared Nothing means that no node or instance is dependent on any resource not on that node or instance. As such, instances cannot access data files owned by another instance. I agree that the terminology "shared disk" is very misleading, as is "Active/Active" and all its variants. Mike Hotek does the rant better than I do on this topic. I always use "multi-connected" disk and emphasize that Clustering arbitrates ownership so only one node actually controls a resource at any given time. -- Geoff N. Hiten Principal SQL Infrastructure Consultant Microsoft SQL Server MVP "Roy Harvey (SQL Server MVP)" <roy_harvey (AT) snet (DOT) net> wrote in message news:5ulab45er6g0o6jh5vuken2466dll25ons (AT) 4ax (DOT) com... On Wed, 27 Aug 2008 09:16:00 +0100, "Edwin vMierlo [MVP]" EdwinvMierlo (AT) discussions (DOT) microsoft.com> wrote: Oh, my understanding was that with active/active that it is load balancing the same database on shared storage and if one fails the other just keeps on trucking. That can't be done? common misunderstanding, Microsoft Failover clustering is a "shared nothnig" cluster model, Microsoft failover clustering is not "shared nothing". The disks are quite clearly shared. The closest SQL Server comes to shared nothing is distributed partitioned views. Roy Harvey Beacon Falls, CT |
#79
| |||
| |||
|
|
Actually a better definition of the 'shared nothing' model is that although resources (disks, IPs, network names, applications, etc.) can move between nodes in a cluster, one node and ONLY one node has access or "owns" those resources at a time. You can see this with disks in that if you try and access a disk in a cluster owned by another node -- Jeff Hughes, MCSE Senior Support Escalation Engineer Microsoft Enterprise Platforms Support (Server Core/Cluster) "Geoff N. Hiten" <SQLCraftsman (AT) gmail (DOT) com> wrote in message news:OuCnY3ECJHA.1628 (AT) TK2MSFTNGP02 (DOT) phx.gbl... Shared Nothing means that no node or instance is dependent on any resource not on that node or instance. As such, instances cannot access data files owned by another instance. I agree that the terminology "shared disk" is very misleading, as is "Active/Active" and all its variants. Mike Hotek does the rant better than I do on this topic. I always use "multi-connected" disk and emphasize that Clustering arbitrates ownership so only one node actually controls a resource at any given time. -- Geoff N. Hiten Principal SQL Infrastructure Consultant Microsoft SQL Server MVP "Roy Harvey (SQL Server MVP)" <roy_harvey (AT) snet (DOT) net> wrote in message news:5ulab45er6g0o6jh5vuken2466dll25ons (AT) 4ax (DOT) com... On Wed, 27 Aug 2008 09:16:00 +0100, "Edwin vMierlo [MVP]" EdwinvMierlo (AT) discussions (DOT) microsoft.com> wrote: Oh, my understanding was that with active/active that it is load balancing the same database on shared storage and if one fails the other just keeps on trucking. That can't be done? common misunderstanding, Microsoft Failover clustering is a "shared nothnig" cluster model, Microsoft failover clustering is not "shared nothing". The disks are quite clearly shared. The closest SQL Server comes to shared nothing is distributed partitioned views. Roy Harvey Beacon Falls, CT |
#80
| |||
| |||
|
|
Actually a better definition of the 'shared nothing' model is that although resources (disks, IPs, network names, applications, etc.) can move between nodes in a cluster, one node and ONLY one node has access or "owns" those resources at a time. You can see this with disks in that if you try and access a disk in a cluster owned by another node -- Jeff Hughes, MCSE Senior Support Escalation Engineer Microsoft Enterprise Platforms Support (Server Core/Cluster) "Geoff N. Hiten" <SQLCraftsman (AT) gmail (DOT) com> wrote in message news:OuCnY3ECJHA.1628 (AT) TK2MSFTNGP02 (DOT) phx.gbl... Shared Nothing means that no node or instance is dependent on any resource not on that node or instance. As such, instances cannot access data files owned by another instance. I agree that the terminology "shared disk" is very misleading, as is "Active/Active" and all its variants. Mike Hotek does the rant better than I do on this topic. I always use "multi-connected" disk and emphasize that Clustering arbitrates ownership so only one node actually controls a resource at any given time. -- Geoff N. Hiten Principal SQL Infrastructure Consultant Microsoft SQL Server MVP "Roy Harvey (SQL Server MVP)" <roy_harvey (AT) snet (DOT) net> wrote in message news:5ulab45er6g0o6jh5vuken2466dll25ons (AT) 4ax (DOT) com... On Wed, 27 Aug 2008 09:16:00 +0100, "Edwin vMierlo [MVP]" EdwinvMierlo (AT) discussions (DOT) microsoft.com> wrote: Oh, my understanding was that with active/active that it is load balancing the same database on shared storage and if one fails the other just keeps on trucking. That can't be done? common misunderstanding, Microsoft Failover clustering is a "shared nothnig" cluster model, Microsoft failover clustering is not "shared nothing". The disks are quite clearly shared. The closest SQL Server comes to shared nothing is distributed partitioned views. Roy Harvey Beacon Falls, CT |
![]() |
| Thread Tools | |
| Display Modes | |
| |