Обсуждение: postgresql cluster on SAN
Hello everyone:
I am a student of the University of Informatics Science in Cuba. I want create a postgresql cluster and I have two servers but I have a doubt. Is possible to create a cluster of postgresql server by sharing the database file on a SAN?
Regards.
Thank you very much for your time.
I am a student of the University of Informatics Science in Cuba. I want create a postgresql cluster and I have two servers but I have a doubt. Is possible to create a cluster of postgresql server by sharing the database file on a SAN?
Regards.
Thank you very much for your time.
On 9/09/2010 10:27 AM, Lazaro Ruben Garcia Martinez wrote: > Hello everyone: > > I am a student of the University of Informatics Science in Cuba. I > want create a postgresql cluster and I have two servers but I have a > doubt. Is possible to create a cluster of postgresql server by sharing > the database file on a SAN? No, PostgreSQL does not support shared-storage clsutering. Use one of the replication-based clustering options like Bucardo or Slony. Added to FAQ after "What replication solutions are available?": http://wiki.postgresql.org/wiki/FAQ#Is_possible_to_create_a_shared-storage_postgresql_server_cluster.3F -- Craig Ringer Tech-related writing at http://soapyfrogs.blogspot.com/
Thank you very much for your answer, In the cluster that i said before I need only failover.
In the documentation of postgresql I read about the Shared Disk Failover, this tecnique avoids synchronization overhead by having only one copy of the database. It uses a single disk array that is shared by multiple servers. If the main database server fails, the standby server is able to mount and start the database as though it was recovering from a database crash. This allows rapid failover with no data loss. One disadvantage is that the standby server should never access the shared storage while the primary server is running.
For these resons is posible to use a SAN?
Regards.
----- "Craig Ringer" <craig@postnewspapers.com.au> escribió:
> On 9/09/2010 10:27 AM, Lazaro Ruben Garcia Martinez wrote:
> > Hello everyone:
> >
> > I am a student of the University of Informatics Science in Cuba. I
> > want create a postgresql cluster and I have two servers but I have a
> > doubt. Is possible to create a cluster of postgresql server by sharing
> > the database file on a SAN?
>
> No, PostgreSQL does not support shared-storage clsutering. Use one of
> the replication-based clustering options like Bucardo or Slony.
>
> Added to FAQ after "What replication solutions are available?":
>
> http://wiki.postgresql.org/wiki/FAQ#Is_possible_to_create_a_shared-storage_postgresql_server_cluster.3F
>
> --
> Craig Ringer
>
> Tech-related writing at http://soapyfrogs.blogspot.com/
>
In the documentation of postgresql I read about the Shared Disk Failover, this tecnique avoids synchronization overhead by having only one copy of the database. It uses a single disk array that is shared by multiple servers. If the main database server fails, the standby server is able to mount and start the database as though it was recovering from a database crash. This allows rapid failover with no data loss. One disadvantage is that the standby server should never access the shared storage while the primary server is running.
For these resons is posible to use a SAN?
Regards.
----- "Craig Ringer" <craig@postnewspapers.com.au> escribió:
> On 9/09/2010 10:27 AM, Lazaro Ruben Garcia Martinez wrote:
> > Hello everyone:
> >
> > I am a student of the University of Informatics Science in Cuba. I
> > want create a postgresql cluster and I have two servers but I have a
> > doubt. Is possible to create a cluster of postgresql server by sharing
> > the database file on a SAN?
>
> No, PostgreSQL does not support shared-storage clsutering. Use one of
> the replication-based clustering options like Bucardo or Slony.
>
> Added to FAQ after "What replication solutions are available?":
>
> http://wiki.postgresql.org/wiki/FAQ#Is_possible_to_create_a_shared-storage_postgresql_server_cluster.3F
>
> --
> Craig Ringer
>
> Tech-related writing at http://soapyfrogs.blogspot.com/
>
On Wed, Sep 8, 2010 at 9:02 PM, Lazaro Ruben Garcia Martinez <lgarciam@estudiantes.uci.cu> wrote: > Thank you very much for your answer, In the cluster that i said before I > need only failover. > In the documentation of postgresql I read about the Shared Disk Failover, > this tecnique avoids synchronization overhead by having only one copy of the > database. It uses a single disk array that is shared by multiple servers. If > the main database server fails, the standby server is able to mount and > start the database as though it was recovering from a database crash. This > allows rapid failover with no data loss. One disadvantage is that the > standby server should never access the shared storage while the primary > server is running. > > For these resons is posible to use a SAN? Yes, however, a SAN is not a replacement for some kind of streaming replication in case you need to recover a database to a previous uncorrupted state or should the SAN fail in some catastrophic way. You need to look up "fencing" for your servers. -- To understand recursion, one must first understand recursion.
On 09/08/10 8:02 PM, Lazaro Ruben Garcia Martinez wrote: > Thank you very much for your answer, In the cluster that i said before > I need only failover. > In the documentation of postgresql I read about the Shared Disk > Failover, this tecnique avoids synchronization overhead by having only > one copy of the database. It uses a single disk array that is shared > by multiple servers. If the main database server fails, the standby > server is able to mount and start the database as though it was > recovering from a database crash. This allows rapid failover with no > data loss. One disadvantage is that the standby server should never > access the shared storage while the primary server is running. > > For these resons is posible to use a SAN? sure. you'll also want cluster management software to coordinate that, such as Heartbeat, Veritas Cluster Service, etc. most vendors who design failover clusters use some sort of storage fencing, such that the standby host *can't* access the shared storage until the 'fencing' is switched. this is often done via port disable on a managed switch (FC or Ethernet).
On 9/09/2010 11:02 AM, Lazaro Ruben Garcia Martinez wrote: > Thank you very much for your answer, In the cluster that i said before I need only failover. > In the documentation of postgresql I read about the Shared Disk Failover, this tecnique avoids synchronization overheadby having only one copy of the database. It uses a single disk array that is shared by multiple servers. If the maindatabase server fails, the standby server is able to mount and start the database as though it was recovering from adatabase crash. This allows rapid failover with no data loss. One disadvantage is that the standby server should never accessthe shared storage while the primary server is running. > For these resons is posible to use a SAN? Yes. See: http://wiki.postgresql.org/wiki/Shared_Storage Setting it up isn't trivial and if you mess it up, you *WILL* get massive data corruption. If your shared storage dies, you still lose all your data. Personally, I favour replication. Anyway, added a reference to [[Shared_Storage]] for failover-only clustering to the faq entry posted earlier. -- Craig Ringer Tech-related writing at http://soapyfrogs.blogspot.com/