Re: is data mirroing online? exchange2003 cluster

From: Leif Pedersen [MVP] (leif.pedersenNOSPAM_at_get2net.dk)
Date: 06/17/04

  • Next message: Gary Massengale: "Hotmail Queue keeps backiing up - reboot server and it clears it Why?" 
    Date: Thu, 17 Jun 2004 22:05:42 +0200

    Hi,

    When a node in a cluster fails the Exchange virtual server will perform a
    failover to the still living node. The disk which contains the databases is
    part of the EVS and will now be connected to the other node and the users
    will be able continue to send/receive emails.

    Leif

    "eric romero" <e.romero@cgnet.com> skrev i en meddelelse
    news:uB1zCQKVEHA.584@TK2MSFTNGP09.phx.gbl...
    > Hi,
    >
    > Yes thanks for your answer, I know A/A is not recommended.
    >
    > I understand the A/P scenario.
    >
    > But my question is still for A/A, if I have nodeA and NodeB (each one has
    > its own databses in its own physical server). what happens if NODEA fails
    > (completely crashed not only the email services)? what happens to the user
    > who connect to NodeA..as far as I understand NodeA databases are not in
    > NodeB (because cluster does not replicate).
    >
    > thx
    >
    >
    > "Matthew Byrd [MSFT]" <matbyrd@online.microsoft.com> wrote in message
    > news:OlN4HaIVEHA.3016@tk2msftngp13.phx.gbl...
    > > Hi Eric,
    > >
    > > First Microsoft does not recommend an A/A configuration as it has
    > > limitations that the A/P configuration does not. These are outline in
    the
    > > cluster doc.
    > >
    > > With that said it appears that some additional information on How
    cluster
    > > works would help to clarify things.
    > >
    > > Windows Cluster does NOT replicate exchange data in any way manner or
    > form.
    > > What you do is setup a set of shared disks that are connected to two
    > > physical Servers. Only ONE of the servers can access a given set of
    > shared
    > > disk at one time. What you gain with a cluster is protection against
    > > hardware failure with the server hardware. You can also get some gains
    in
    > > doing updates and maintenance to the software as you can do the
    > maintenance
    > > on the passive node then fail over.
    > >
    > > So in an A/P configuration this is how a failover would occur.
    > >
    > > Node A is currently active and running Exchange
    > > A hardware Failure occurs on Node A
    > > Node B determines that Node A has gone offline
    > > Node B takes over the shared disks and starts brining Exchange back
    online
    > > On Node B Exchange Does a soft recovery and brings the (Exact same)
    > database
    > > online
    > > Users are now able to connect again
    > >
    > > This scenario is EXACTLY the same in an A/A configuration except that
    > after
    > > the failover there would be two Exchange Virtual servers running on the
    > same
    > > physical Box. Each Exchange Virtual Server in an A/A configuration has
    > its
    > > own databases, its own separate shared disks, and its own set of users.
    > >
    > > Hopefully this has addressed your question
    > >
    > > --
    > > Matthew Byrd
    > > Microsoft PSS
    > >
    > > When responding to posts, please "Reply to Group" via your newsreader so
    > > that others may learn and benefit from your issue.
    > >
    > > This posting is provided "AS IS" with no warranties, and confers no
    > rights.
    > >
    > >
    > >
    > >
    > > "eric romero" <e.romero@cgnet.com> wrote in message
    > > news:eT%23uEV%23UEHA.2972@TK2MSFTNGP11.phx.gbl...
    > > > Hi Matthew,
    > > >
    > > > That document does not explain what do they do to keep exchange
    > database
    > > > synchronized in an active/active configuration.
    > > >
    > > > Do you know if in active/active "some mechanism" replicates "all the
    > time"
    > > > data from NodeA to NodeB, so in the vent of NodeA going offline then
    > NodeB
    > > > takes place? and also I do not understand why do they alert on number
    of
    > > > storage per server..is it that NodeB and NodeA can have diffente
    number
    > of
    > > > staorages? should not they be two identical servers/databases/paths.
    > > >
    > > > thx
    > > >
    > > > "Matthew Byrd [MSFT]" <matbyrd@online.microsoft.com> wrote in message
    > > > news:ODnVUr9UEHA.2816@TK2MSFTNGP11.phx.gbl...
    > > >> Hi Eric,
    > > >>
    > > >> From your post it appears that you are looking for an Exchange High
    > > >> Availability solution but that you are not clear on what clustering
    > does
    > > > to
    > > >> assist that.
    > > >>
    > > >> I would recommend taking a look at the following whitepaper I think
    it
    > > > will
    > > >> answer many of your High Availability questions:
    > > >>
    > > >> Exchange Server 2003 High Availability Guide
    > > >>
    > > >
    >
    http://www.microsoft.com/technet/prodtechnol/exchange/2003/library/highavailgde.mspx
    > > >>
    > > >> Hope this Helps,
    > > >>
    > > >> --
    > > >> Matthew Byrd
    > > >> Microsoft PSS
    > > >>
    > > >> When responding to posts, please "Reply to Group" via your newsreader
    > so
    > > >> that others may learn and benefit from your issue.
    > > >>
    > > >> This posting is provided "AS IS" with no warranties, and confers no
    > > > rights.
    > > >>
    > > >> "eric romero" <e.romero@cgnet.com> wrote in message
    > > >> news:%23H8k1B9UEHA.2544@TK2MSFTNGP10.phx.gbl...
    > > >> > Hi all,
    > > >> >
    > > >> > Could you please tell me and show me some documents that explains
    > what
    > > >> > happens with the exchange databases on an Exchange2003 cluster (2
    > > >> > nodes)
    > > >> > config?
    > > >> > I am interested to know if we can have 2 nodes each node with its
    own
    > > > hard
    > > >> > disks and data from active server is copying online to the standby
    > > > server
    > > >> > so
    > > >> > it is ready to serve the clients if the active node fails.
    > > >> >
    > > >> > Is it true that on active/pasive I can have just 1 place to install
    > the
    > > >> > exchange databases, what happens if this "shared disk" fails? what
    is
    > > > the
    > > >> > cluster useful for at thismoment?
    > > >> > on active/active I think I can have both server available at any
    > time,
    > > > how
    > > >> > are the exchange databases synchronized? or is it in active/active
    > that
    > > >> > you
    > > >> > set some mailboxes to be serverrd by node1 and the rest by node2,
    is
    > > > this
    > > >> > is
    > > >> > the case what if one of the nodes fail?
    > > >> >
    > > >> > thx
    > > >> >
    > > >> >
    > > >>
    > > >>
    > > >
    > > >
    > >
    > >
    > >
    >
    >

  • Next message: Gary Massengale: "Hotmail Queue keeps backiing up - reboot server and it clears it Why?"

    Relevant Pages

    • Re: is data mirroing online? exchange2003 cluster
      ... The disk which contains the databases is ... > its own databses in its own physical server). ... > NodeB (because cluster does not replicate). ... >> What you do is setup a set of shared disks that are connected to two ...
      (microsoft.public.exchange2000.setup.installation)
    • Re: is data mirroing online? exchange2003 cluster
      ... The disk which contains the databases is ... > its own databses in its own physical server). ... > NodeB (because cluster does not replicate). ... >> What you do is setup a set of shared disks that are connected to two ...
      (microsoft.public.exchange2000.admin)
    • Windows Server 2003 Cluster iSCSI drives missing from standby node
      ... I have 2 x Windows Server 2003 servers as SQL Cluster. ... to take load as the disks are not appearing in Disk Management or iSCSI ... The same setting is on Active node SRV-A and which is able to see the iSCSI ...
      (microsoft.public.windows.server.clustering)
    • RE: Clustering a standalone SQL server + storage array
      ... working as expected with out error your ready to install sql server. ... log files to the data and log shared drives you setup as part of the cluster. ... Our apps team look after a trading system that uses an HP DL360 server ... running SQL 2000 SP3a with a few small databases on it. ...
      (microsoft.public.sqlserver.clustering)
    • Re: upgrading sql2k cluster to sql2005
      ... It ain't that simple for SQL Server. ... that databases get locked off in the current cluster and then properly ...
      (microsoft.public.sqlserver.clustering)
    Loading