"Danny Sanders" <Danny.Sanders@NO-SPAMcpcmed.org> wrote in message
news:OTxcU0pUGHA.4900@TK2MSFTNGP12.phx.gbl...
> I'm looking for an explanation for why companies use a "front end" and
> "back end" Exchange server setup. Is it for redundancy? If so what
> redundancy does it provide?
>
> Why would a company use this set up as opposed to just a single server?
>
> Looking for ways to provide redundancy for an Exchange server without
> including clustering. We have IBM CommonStore for Exchange, which is an
> email archive system, and I think the combination of the email archive,
> Exchange, and a Windows cluster would be introducing an undue amount of
> complexity to the network. I'm willing to look at clustering but I want
> that to be a last resort.
>
> I'm hoping there are ways to  provide redundancy and "KISS".
>
>
> TIA
> DDS
>

"Bharat Suneja" <bharatsuneja@no.spam.org> wrote in message
news:uChqWsqUGHA.1576@tk2msftngp13.phx.gbl...
>- Front-Ends are used for protocol virtual servers - typically for OWA
>(HTTP/HTTPS), but also for IMAP and POP3.
> - All Exchange servers are "back-ends" by default - it's not an explicit
> designation/server role. If you make a server a "front-end" it's not a
> "back-end", else all servers are "back-ends" by default.
>
> What do Front-Ends buy you?
> - Simplification of OWA/IMAP/POP3 URLs in case of multiple back-end
> Exchange servers. These users will simply need to connect to the
> Front-End. Users don't need to connect to Backend1 or Backend2... which
> would be the case without a front-end.
> - Redundancy: Can also use Load-Balancing so it's a hostname (A record)
> mapped to a virtual IP address served by more than one Front-Ends. Users
> don't need to connect to FrontEnd2 when FrontEnd1 goes down.
> - Security: You're opening relevant ports for HTTP/HTTPS, IMAP4, POP3, et
> al, to the Front-Ends only
> - Off-loading SSL from Back-end servers
> - Off-loading protocol processing from Back-end servers
>
> Note, SMTP is not really "front-ended" like the other protocols.
>
> Front-End and Back-End Topology Advantages
> http://www.microsoft.com/technet/prodtechnol/exchange/guides/E2k3FrontBack/e7fa5dcc-d93f-4441-aace-c52e5baedc2a.mspx?mfr=true
>
> - Redundancy of back-end mailbox servers is provided by MSCS/High
> Availability clusters (don't confuse them with NLB or Network
> Load-Balancing - the term "cluster" is also used for NLB).
> - Even with HA clusters, the shared storage is single point of failure -
> it that blows you lose your Stores (amongst other stuff located on shared
> storage like mailroot... ). Replication products from storage vendors like
> NetApp (SnapMirror) & EMC is one way to mitigate that risk, but each piece
> you add on top increases complexity of your environment as also the cost.
> --
> Bharat Suneja
> MCSE, MCT
> www.zenprise.com
> blog: www.suneja.com/blog
> -----------------------------------------
>
>
> "Danny Sanders" <Danny.Sanders@NO-SPAMcpcmed.org> wrote in message
> news:OTxcU0pUGHA.4900@TK2MSFTNGP12.phx.gbl...
>> I'm looking for an explanation for why companies use a "front end" and
>> "back end" Exchange server setup. Is it for redundancy? If so what
>> redundancy does it provide?
>>
>> Why would a company use this set up as opposed to just a single server?
>>
>> Looking for ways to provide redundancy for an Exchange server without
>> including clustering. We have IBM CommonStore for Exchange, which is an
>> email archive system, and I think the combination of the email archive,
>> Exchange, and a Windows cluster would be introducing an undue amount of
>> complexity to the network. I'm willing to look at clustering but I want
>> that to be a last resort.
>>
>> I'm hoping there are ways to  provide redundancy and "KISS".
>>
>>
>> TIA
>> DDS
>>
>
>

>
>
> DDS
> "Bharat Suneja" <bharatsuneja@no.spam.org> wrote in message
> news:uChqWsqUGHA.1576@tk2msftngp13.phx.gbl...
>>- Front-Ends are used for protocol virtual servers - typically for OWA
>>(HTTP/HTTPS), but also for IMAP and POP3.
>> - All Exchange servers are "back-ends" by default - it's not an explicit
>> designation/server role. If you make a server a "front-end" it's not a
>> "back-end", else all servers are "back-ends" by default.
>>
>> What do Front-Ends buy you?
>> - Simplification of OWA/IMAP/POP3 URLs in case of multiple back-end
>> Exchange servers. These users will simply need to connect to the
>> Front-End. Users don't need to connect to Backend1 or Backend2... which
>> would be the case without a front-end.
>> - Redundancy: Can also use Load-Balancing so it's a hostname (A record)
>> mapped to a virtual IP address served by more than one Front-Ends. Users
>> don't need to connect to FrontEnd2 when FrontEnd1 goes down.
>> - Security: You're opening relevant ports for HTTP/HTTPS, IMAP4, POP3, et
>> al, to the Front-Ends only
>> - Off-loading SSL from Back-end servers
>> - Off-loading protocol processing from Back-end servers
>>
>> Note, SMTP is not really "front-ended" like the other protocols.
>>
>> Front-End and Back-End Topology Advantages
>> http://www.microsoft.com/technet/prodtechnol/exchange/guides/E2k3FrontBack/e7fa5dcc-d93f-4441-aace-c52e5baedc2a.mspx?mfr=true
>>
>> - Redundancy of back-end mailbox servers is provided by MSCS/High
>> Availability clusters (don't confuse them with NLB or Network
>> Load-Balancing - the term "cluster" is also used for NLB).
>> - Even with HA clusters, the shared storage is single point of failure -
>> it that blows you lose your Stores (amongst other stuff located on shared
>> storage like mailroot... ). Replication products from storage vendors
>> like NetApp (SnapMirror) & EMC is one way to mitigate that risk, but each
>> piece you add on top increases complexity of your environment as also the
>> cost.
>> --
>> Bharat Suneja
>> MCSE, MCT
>> www.zenprise.com
>> blog: www.suneja.com/blog
>> -----------------------------------------
>>
>>
>> "Danny Sanders" <Danny.Sanders@NO-SPAMcpcmed.org> wrote in message
>> news:OTxcU0pUGHA.4900@TK2MSFTNGP12.phx.gbl...
>>> I'm looking for an explanation for why companies use a "front end" and
>>> "back end" Exchange server setup. Is it for redundancy? If so what
>>> redundancy does it provide?
>>>
>>> Why would a company use this set up as opposed to just a single server?
>>>
>>> Looking for ways to provide redundancy for an Exchange server without
>>> including clustering. We have IBM CommonStore for Exchange, which is an
>>> email archive system, and I think the combination of the email archive,
>>> Exchange, and a Windows cluster would be introducing an undue amount of
>>> complexity to the network. I'm willing to look at clustering but I want
>>> that to be a last resort.
>>>
>>> I'm hoping there are ways to  provide redundancy and "KISS".
>>>
>>>
>>> TIA
>>> DDS
>>>
>>
>>
>
>

"Bharat Suneja" <bharatsuneja@no.spam.org> wrote in message
news:uaCsoprUGHA.5660@TK2MSFTNGP12.phx.gbl...
> Inline.
>
> --
> Bharat Suneja
> MCSE, MCT
> www.zenprise.com
> blog: www.suneja.com/blog
> -----------------------------------------
>
>
> "Danny Sanders" <Danny.Sanders@NO-SPAMcpcmed.org> wrote in message
> news:eL9tZhrUGHA.6112@TK2MSFTNGP10.phx.gbl...
>> Thanks for that information.
>>
>>> - Even with HA clusters, the shared storage is single point of failure -
>>> it that blows you lose your Stores (amongst other stuff located on
>>> shared storage like mailroot... )
>>
>> Question:
>>
>> With a front end server and two backend servers, would there be two
>> exchange stores on the backend servers?
> Yes, 2 or more. Exchange 200x Stores consist of 2 files, the EDB & STM.
> Together they make up one database.
> Databases are organized into Storage Groups. By default you will have the
> first storage group on each back-end with a public and a private (mailbox)
> store. With Enterprise Edition you can have up to 4 Storage Groups with up
> to 5 Stores in each.
>
>> If so would those stores be identical?
> Starting out, they will be empty stores. Each server will have the system
> mailboxes on it.
>
>>What I would like to accomplish is to have two identical exchange stores
>>that are created simultaneously as opposed to "replicated".
> Not sure what that'd achieve. Do you want both stores to have the same
> data (same mailboxes replicated to each Store)? If yes, that's not
> possible with any Exchange-native feature/tool.
>
>> I had a situation where hardware corrupted a store and I got a perfect
>> backup of the useless corrupt store.
> Which is why you have multiple backups so you can always go to a previous
> day's backup.
>
>>
>> Is there a way to do this?
> Stores could be "replicated" (yes, we're back to replication because
> that's what I think you want to get... 2 mailbox stores with the same data
> so if one goes down you can use the other one) using 3rd party utilities
> like DoubleTake. Alternatively, you can wait for the next version of
> Exchange ("12") to ship, which is reported to have log shipping that'll
> provide a solution for this requirement.
>
>>
>>
>> DDS
>> "Bharat Suneja" <bharatsuneja@no.spam.org> wrote in message
>> news:uChqWsqUGHA.1576@tk2msftngp13.phx.gbl...
>>>- Front-Ends are used for protocol virtual servers - typically for OWA
>>>(HTTP/HTTPS), but also for IMAP and POP3.
>>> - All Exchange servers are "back-ends" by default - it's not an explicit
>>> designation/server role. If you make a server a "front-end" it's not a
>>> "back-end", else all servers are "back-ends" by default.
>>>
>>> What do Front-Ends buy you?
>>> - Simplification of OWA/IMAP/POP3 URLs in case of multiple back-end
>>> Exchange servers. These users will simply need to connect to the
>>> Front-End. Users don't need to connect to Backend1 or Backend2... which
>>> would be the case without a front-end.
>>> - Redundancy: Can also use Load-Balancing so it's a hostname (A record)
>>> mapped to a virtual IP address served by more than one Front-Ends. Users
>>> don't need to connect to FrontEnd2 when FrontEnd1 goes down.
>>> - Security: You're opening relevant ports for HTTP/HTTPS, IMAP4, POP3,
>>> et al, to the Front-Ends only
>>> - Off-loading SSL from Back-end servers
>>> - Off-loading protocol processing from Back-end servers
>>>
>>> Note, SMTP is not really "front-ended" like the other protocols.
>>>
>>> Front-End and Back-End Topology Advantages
>>> http://www.microsoft.com/technet/prodtechnol/exchange/guides/E2k3FrontBack/e7fa5dcc-d93f-4441-aace-c52e5baedc2a.mspx?mfr=true
>>>
>>> - Redundancy of back-end mailbox servers is provided by MSCS/High
>>> Availability clusters (don't confuse them with NLB or Network
>>> Load-Balancing - the term "cluster" is also used for NLB).
>>> - Even with HA clusters, the shared storage is single point of failure -
>>> it that blows you lose your Stores (amongst other stuff located on
>>> shared storage like mailroot... ). Replication products from storage
>>> vendors like NetApp (SnapMirror) & EMC is one way to mitigate that risk,
>>> but each piece you add on top increases complexity of your environment
>>> as also the cost.
>>> --
>>> Bharat Suneja
>>> MCSE, MCT
>>> www.zenprise.com
>>> blog: www.suneja.com/blog
>>> -----------------------------------------
>>>
>>>
>>> "Danny Sanders" <Danny.Sanders@NO-SPAMcpcmed.org> wrote in message
>>> news:OTxcU0pUGHA.4900@TK2MSFTNGP12.phx.gbl...
>>>> I'm looking for an explanation for why companies use a "front end" and
>>>> "back end" Exchange server setup. Is it for redundancy? If so what
>>>> redundancy does it provide?
>>>>
>>>> Why would a company use this set up as opposed to just a single server?
>>>>
>>>> Looking for ways to provide redundancy for an Exchange server without
>>>> including clustering. We have IBM CommonStore for Exchange, which is an
>>>> email archive system, and I think the combination of the email archive,
>>>> Exchange, and a Windows cluster would be introducing an undue amount of
>>>> complexity to the network. I'm willing to look at clustering but I want
>>>> that to be a last resort.
>>>>
>>>> I'm hoping there are ways to  provide redundancy and "KISS".
>>>>
>>>>
>>>> TIA
>>>> DDS
>>>>
>>>
>>>
>>
>>
>
>

"Danny Sanders" <Danny.Sanders@NO-SPAMcpcmed.org> wrote in message
news:uZgl8B3UGHA.4952@TK2MSFTNGP09.phx.gbl...
> Thanks for all the information.
>
>>>What I would like to accomplish is to have two identical exchange stores
>>>that are created simultaneously as opposed to "replicated".
>> Not sure what that'd achieve. Do you want both stores to have the same
>> data (same mailboxes replicated to each Store)? If yes, that's not
>> possible with any Exchange-native feature/tool.
>
>
> I had hardware that caused corruption to the store and for a few weeks we
> were backing up corrupt data. Just trying to see if there is a practical
> way to create two separate exchange stores without replication. If one
> store somehow got corrupt I would not want that corruption replicated.
> It sounds like there is no way to have two identical exchange stores
> without replication (from a third party app?) being involved right?
>
> Thanks again
>
> DDS
> "Bharat Suneja" <bharatsuneja@no.spam.org> wrote in message
> news:uaCsoprUGHA.5660@TK2MSFTNGP12.phx.gbl...
>> Inline.
>>
>> --
>> Bharat Suneja
>> MCSE, MCT
>> www.zenprise.com
>> blog: www.suneja.com/blog
>> -----------------------------------------
>>
>>
>> "Danny Sanders" <Danny.Sanders@NO-SPAMcpcmed.org> wrote in message
>> news:eL9tZhrUGHA.6112@TK2MSFTNGP10.phx.gbl...
>>> Thanks for that information.
>>>
>>>> - Even with HA clusters, the shared storage is single point of
>>>> failure - it that blows you lose your Stores (amongst other stuff
>>>> located on shared storage like mailroot... )
>>>
>>> Question:
>>>
>>> With a front end server and two backend servers, would there be two
>>> exchange stores on the backend servers?
>> Yes, 2 or more. Exchange 200x Stores consist of 2 files, the EDB & STM.
>> Together they make up one database.
>> Databases are organized into Storage Groups. By default you will have the
>> first storage group on each back-end with a public and a private
>> (mailbox) store. With Enterprise Edition you can have up to 4 Storage
>> Groups with up to 5 Stores in each.
>>
>>> If so would those stores be identical?
>> Starting out, they will be empty stores. Each server will have the system
>> mailboxes on it.
>>
>>>What I would like to accomplish is to have two identical exchange stores
>>>that are created simultaneously as opposed to "replicated".
>> Not sure what that'd achieve. Do you want both stores to have the same
>> data (same mailboxes replicated to each Store)? If yes, that's not
>> possible with any Exchange-native feature/tool.
>>
>>> I had a situation where hardware corrupted a store and I got a perfect
>>> backup of the useless corrupt store.
>> Which is why you have multiple backups so you can always go to a previous
>> day's backup.
>>
>>>
>>> Is there a way to do this?
>> Stores could be "replicated" (yes, we're back to replication because
>> that's what I think you want to get... 2 mailbox stores with the same
>> data so if one goes down you can use the other one) using 3rd party
>> utilities like DoubleTake. Alternatively, you can wait for the next
>> version of Exchange ("12") to ship, which is reported to have log
>> shipping that'll provide a solution for this requirement.
>>
>>>
>>>
>>> DDS
>>> "Bharat Suneja" <bharatsuneja@no.spam.org> wrote in message
>>> news:uChqWsqUGHA.1576@tk2msftngp13.phx.gbl...
>>>>- Front-Ends are used for protocol virtual servers - typically for OWA
>>>>(HTTP/HTTPS), but also for IMAP and POP3.
>>>> - All Exchange servers are "back-ends" by default - it's not an
>>>> explicit designation/server role. If you make a server a "front-end"
>>>> it's not a "back-end", else all servers are "back-ends" by default.
>>>>
>>>> What do Front-Ends buy you?
>>>> - Simplification of OWA/IMAP/POP3 URLs in case of multiple back-end
>>>> Exchange servers. These users will simply need to connect to the
>>>> Front-End. Users don't need to connect to Backend1 or Backend2... which
>>>> would be the case without a front-end.
>>>> - Redundancy: Can also use Load-Balancing so it's a hostname (A record)
>>>> mapped to a virtual IP address served by more than one Front-Ends.
>>>> Users don't need to connect to FrontEnd2 when FrontEnd1 goes down.
>>>> - Security: You're opening relevant ports for HTTP/HTTPS, IMAP4, POP3,
>>>> et al, to the Front-Ends only
>>>> - Off-loading SSL from Back-end servers
>>>> - Off-loading protocol processing from Back-end servers
>>>>
>>>> Note, SMTP is not really "front-ended" like the other protocols.
>>>>
>>>> Front-End and Back-End Topology Advantages
>>>> http://www.microsoft.com/technet/prodtechnol/exchange/guides/E2k3FrontBack/e7fa5dcc-d93f-4441-aace-c52e5baedc2a.mspx?mfr=true
>>>>
>>>> - Redundancy of back-end mailbox servers is provided by MSCS/High
>>>> Availability clusters (don't confuse them with NLB or Network
>>>> Load-Balancing - the term "cluster" is also used for NLB).
>>>> - Even with HA clusters, the shared storage is single point of
>>>> failure - it that blows you lose your Stores (amongst other stuff
>>>> located on shared storage like mailroot... ). Replication products from
>>>> storage vendors like NetApp (SnapMirror) & EMC is one way to mitigate
>>>> that risk, but each piece you add on top increases complexity of your
>>>> environment as also the cost.
>>>> --
>>>> Bharat Suneja
>>>> MCSE, MCT
>>>> www.zenprise.com
>>>> blog: www.suneja.com/blog
>>>> -----------------------------------------
>>>>
>>>>
>>>> "Danny Sanders" <Danny.Sanders@NO-SPAMcpcmed.org> wrote in message
>>>> news:OTxcU0pUGHA.4900@TK2MSFTNGP12.phx.gbl...
>>>>> I'm looking for an explanation for why companies use a "front end" and
>>>>> "back end" Exchange server setup. Is it for redundancy? If so what
>>>>> redundancy does it provide?
>>>>>
>>>>> Why would a company use this set up as opposed to just a single
>>>>> server?
>>>>>
>>>>> Looking for ways to provide redundancy for an Exchange server without
>>>>> including clustering. We have IBM CommonStore for Exchange, which is
>>>>> an email archive system, and I think the combination of the email
>>>>> archive, Exchange, and a Windows cluster would be introducing an undue
>>>>> amount of complexity to the network. I'm willing to look at clustering
>>>>> but I want that to be a last resort.
>>>>>
>>>>> I'm hoping there are ways to  provide redundancy and "KISS".
>>>>>
>>>>>
>>>>> TIA
>>>>> DDS
>>>>>
>>>>
>>>>
>>>
>>>
>>
>>
>
>