Re: I have question regarding active/active clustering

With log shipping or database mirroring you have two complete copies of the
databases and the copies can be separated by thousands of miles for
disaster-recovery scenarios. With a cluster, two different servers are
connected to the same physical disk subsystem so there is only one copy of
the data. Only one server is connected to the data files at a time. when
one fails, the other server takes over and continues to use the same data
disks as the original. Clustering is usually done with both servers in the
same room because they have to be connected to disk array or SAN. You can
use a combination of both - failover clustering for hardware failure
recovery and mirroring or log shipping for disaster recovery when the whole
cluster loses power for example. It's possible to lose data after a
log-shipping failover. The cluster won't lose data on failover unless the
disk subsystem fails (which is why clusters usually use a SAN for disk
reliability and redundancy).

--
This posting is provided "AS IS" with no warranties, and confers no rights.
Use of included script samples are subject to the terms specified at
http://www.microsoft.com/info/cpyright.htm

"Iter" <Iter@xxxxxxxxxxxxxxxxxxxxxxxxx> wrote in message
news:E0DB8E1B-A557-42EB-9216-722A872AA503@xxxxxxxxxxxxxxxx
Wow, thank you guys. But other question appeared. What is difference
between
log shipping standby and active/passive cluster? is there any difference
for
the functionality? Thanks

"Roger Wolter[MSFT]" wrote:

After reading through this, I think the issues is you are still thinking
of
a cluster as supporting a single database. An active-active cluster is
in
fact two different active-passive clusters set up so that when both
servers
are operational there are two database instances running on the two
servers.
For example, you might have the payroll database on one server and the
order
entry database on the other server. If one of the servers fails then
both
payroll and order entry are on the same server until the other server is
repaired.

I know this is just a repeat of what Edwin has said so eloquently but I
hope
a more concrete example might clarify things.

--
This posting is provided "AS IS" with no warranties, and confers no
rights.
Use of included script samples are subject to the terms specified at
http://www.microsoft.com/info/cpyright.htm

"Edwin vMierlo" <EdwinvMierlo@xxxxxxxxxxxxxxxxxxxxxxxxx> wrote in message
news:OzF1rqRJHHA.816@xxxxxxxxxxxxxxxxxxxxxxx
in line

"Iter" <Iter@xxxxxxxxxxxxxxxxxxxxxxxxx> wrote in message
news:F1E612C0-8D84-4967-B113-E8690429E00D@xxxxxxxxxxxxxxxx
Thank you so much, edwin, your answer is very helpful. I have other
question
related to it. for active/active, Is it used log shipping same as
active/passive?

clustering has nothing to do with log-shipping. So think about if you
have
two seperate standalone servers, and you want to do log shipping, you
can.
The same goes for 2 SQL instances in a cluster, if you want to do
log-shipping you can.

Does one node take over other node resource when failed? I
unstand it is not real time, but is it fail over?

Yes, If a node fails, everything running (under cluster) on that node
is
going to run on the other surviving node. so Yes, your SQL instance
will
run
on the other node. The surviving node is now running 2 SQL instances,
but
still they are considered to be two *seperate servers*.
It is not real time, there is a period where the SQL instance which was
running on the other node is offline. This period of time can be very
short,
matter of seconds, or somewhat longer, matter of minutes. The period it
is
offline is influenced by how you set up your cluster group, how many
resources, how many disks, and so on, so forth.

I hope this helps,
Rgds,
Edwin.

Thanks

"Edwin vMierlo" wrote:

in-line

"Iter" <Iter@xxxxxxxxxxxxxxxxxxxxxxxxx> wrote in message
news:DC6C50FC-DF66-4701-806A-D33B366FBE65@xxxxxxxxxxxxxxxx
I still got confused. From your explaination and that web site you
gave to
me, I think active/active cluster is like RAC in oracle, that
means
they
can
share the resource

no

or load banlance

no

and a node can take over other node that
is failed when it is two nodes cluster.

yes

Active/passive cluster is like oracle
standby, right?

I think so, but I don't know oracle


Is active/active a real product or just a theory?

Active/Active is a configuration of failover clustering. You can run
one
Virtual SQL Instance (think of an instance as a standalone server)
on
one
node, and run another Virtual SQL Instance on a second node. On two
nodes
you are now running two SQL instances. Effectively the two instances
are
exactly the same as running two seperate standalone SQL servers. The
do
not
share any resource, they do not load balance in anyway, each having
a
different name, each having a different IP, each having there own
databases.

does this explain it better to you ?
rgds,
Edwin.



Thanks.



"Edwin vMierlo" wrote:

active / passive for SQL server for a two node cluster means you
have
only
one group with an SQL instance. Only one node is running this
SQL
instance,
the other node is not running the SQL instance and is *standby*
to
take
ownership if a fail condition occurs on the node running.

active / active for SQL server for a two node cluster menas you
have
more
than one group with SQL instances. All nodes in your cluster run
(at
least)
one SQL instance. In case of a fail condition, all groups will
go
to
the
surviving node, and this node will run all SQL instances.
Therefore
this
might have an impact on Performance, in such a setup you need to
ensure
you
scale your hardware to be able to take all resources and still
have
acceptable performance.

There are many more definitions of active / passive and active
/active,
and
some of them are confusing, and different people give different
meaning
to
these terms. Probably better to talk about :
- 2 node cluster with 1 SQL instance or
- 2 node cluster with multiple SQL instances

For a Microsoft definition of active / passive and active /
active,
please
see :


http://technet2.microsoft.com/WindowsServer/en/library/8846a72b-0882-4a24-8eee-a768e52925281033.mspx?mfr=true
(scroll down to the question "What is active/active verses
active/passive?")

HTH,
_Edwin.

"Iter" <Iter@xxxxxxxxxxxxxxxxxxxxxxxxx> wrote in message
news:84780791-D41C-4482-8E59-6C799157C32C@xxxxxxxxxxxxxxxx
That means it should equivelant to standby in oracle right?
also,
what
is
different betteen active/pasive and active/active? In
active/active,
does
it
use log shipping same as active/pasive? Thanks.



"Edwin vMierlo" wrote:

Iter,

The answer is No (unfortunately)

Failover clustering is not like oracle RAC, an SQL Instance
contains
its
own
databases and only runs on one of the nodes in your cluster.
In
case
of
a
failover, there will be a disconnect of connections as
cluster
will
offline
the group (and therefore the applications/SQL) while moving
the
ownership to
another node in the cluster.

You cannot have 2 instances running off the same
databases/disks.
Failover
Clustering is a "shared nothing" cluster model.

rgds,
Edwin.




"Iter" <Iter@xxxxxxxxxxxxxxxxxxxxxxxxx> wrote in message
news:3E3EFAA9-84C3-40F9-83F8-7EA5CAECA99B@xxxxxxxxxxxxxxxx
Hi Guys,
I have question regarding active/active clustering. Is
that
this
mode
can
use for load balance and server redundance that means the
connection
keep
alive when one of node failed. Is it like oracle RAC?
Thanks.

















.

Relevant Pages

  • Re: I have question regarding active/active clustering
    ... "Edwin vMierlo" wrote: ... two seperate standalone servers, and you want to do log shipping, you can. ... Yes, If a node fails, everything running (under cluster) on that node is ... there is a period where the SQL instance which was ...
    (microsoft.public.sqlserver.clustering)
  • Globally separated cluster
    ... I am trying to plan a SQL Server 2005 cluster across multiple data ... Basically the database backs a load balanced ... Each site has its own set of web servers and its own ...
    (microsoft.public.sqlserver.clustering)
  • Re: Windows 2003 Network Load Balancing Problem
    ... Sadly my servers websites were configured with specific IPs and host ... I find it very strange why the NLB driver can receive on a virtual IP ... he set his IIS website to use ALL UNASSIGNED addresses rather than ... specifically pointing it at the single virtual cluster address. ...
    (microsoft.public.windows.server.clustering)
  • Re: upgrading sql2k cluster to sql2005
    ... that databases get locked off in the current cluster and then properly ... I maintain many SQL servers as ... Who is going to track down all of the applications that need to have ...
    (microsoft.public.sqlserver.clustering)
  • Re: Load Balanced TS Servers and Unicast Flooding
    ... Remove each on of your tse servers from the cluster, ... Start adding them back to the NLB Cluster. ... > NICs. ...
    (microsoft.public.windows.terminal_services)