|
|
|
date: Wed, 27 Feb 2008 13:14:03 -0800,
group: microsoft.public.exchange.clustering
back
Re: Single SCR server to service multiple Exchange 2007 servers
Thanks for the followup info and the kind words.
What messaging system are you using now, and what sort of downtime have you
experienced in the past? There's so much more that can go wrong than just a
server being down. What about storage or other hardware failures? What
about facilities failures (power, cooling, etc.)? What about the biggest
cause of failures of all - human error?
Sounds like from your post that you're the admin in charge of designing and
implementing the solution. Part of your job (as you probably know) is to
help educate the decision makers and set the expectations on the services
provided by the solution (e.g., availability, uptime, etc., etc.). The core
tenet of Exchange HA is to introduce redundancy throughout the environment.
And it is the business needs that drive the design of the overall HA
solution.
In your organization, it sounds like you're saying the decision makers have
decided that, although you have a lot of mailboxes, access to mail data or
the ability to send/receive messages is not that critical for your users.
If you're running Exchange now (e.g., Exchange 2003, etc.) then you may be
able to get good usage statistics by collection a variety of performance
counters. If you're not running Exchange, there still might be counters you
can collect. Of course, if you have no messaging system now, then you might
not know readily know what the specific availability needs are and will have
to wait and see.
BTW, one comment on your design of 3-4 servers and using CCR. When you
deploy CCR, that will be two of your servers. That would leave you one or
two servers for the Client Access and Hub Transport roles. If you only have
one server for CAS/Hub, then you've got a bad single point of failure. If
that server goes down you lose mail flow, and much, if not all, client
access.
--
Regards,
Scott Schnoll
Microsoft Corporation
This posting is provided "AS IS" with no warranties, and confers no
rights. Please do not send email directly to this alias. This alias is for
newsgroup purposes only.
"Jon" wrote in message
news:27025D8D-54EE-4BD4-8875-DA83295927C6@microsoft.com...
>I understand where you are coming from but the organization I work for, as
> large as they are, does not have much of that kind of information
> available
> and they don't seem overly concerned by the idea of an Exchange server
> being
> down for a while (although that story will change when the time comes I'm
> afraid). As it sits now they do not want to spend any extra money on
> redundancy or fault tolerance and this was my idea of something that I
> could
> attempt to sell them on where it would be the lowest cost to have some
> type
> of "poor man's clustering" in our environment and cover all our
> servers/databases. I am not accustomed to being in an environment where
> they
> do not deem Exchange to be critical enough to put money into good
> redundancy
> configurations. It's been a struggle so far.
>
> I actually just watched your demos on the Exchange blog site about an hour
> before I posted this so it was funny to see you reply to this. Great job
> with that. The Exchange blog site is the best site ever for Exchange
> admins.
> Thanks for all the info that you all contribute there.
>
> "Scott Schnoll [MSFT]" wrote:
>
>> Let's first step back and focus on the business requirements. That
>> really
>> should be what drives the solution architecture and design.
>>
>> What sort of SLAs do you have for the messaging service and data
>> availability?
>>
>> What are your RTO and RPO?
>> --
>> Regards,
>>
>> Scott Schnoll
>> Microsoft Corporation
>> This posting is provided "AS IS" with no warranties, and confers no
>> rights. Please do not send email directly to this alias. This alias is
>> for
>> newsgroup purposes only.
>>
>>
>> "Jon" wrote in message
>> news:CFE8BCD1-FF48-445E-B948-2836E8AD5F61@microsoft.com...
>> >I have an enviroment that is looking to have around 3 or 4 Exchange 2007
>> > servers and 10,000 users. I would like to do CCR for all the servers
>> > but
>> > cannot convince the powers that be with the money to spend that much on
>> > additional hardware and software. I was wondering if one bulky SCR
>> > server
>> > could handle 3 or 4 Exchange 2007 servers constantly pushing their logs
>> > and
>> > data to it. The idea is we can have this SCR server and use database
>> > portability in the case of a disaster and this one box could support
>> > being
>> > the "clustered" server for all 3 or 4 other servers. I realize this is
>> > not a
>> > cluster and it would require manual steps to bring a database up if
>> > necessary. I know every environment is different and a lot of this
>> > depends
>> > on how much mail flow we have and how large our databases are etc. I
>> > just
>> > wondered if using one SCR instance and sharing it for multiple live
>> > Exchange
>> > 2007 servers is something people have done and if they have had good
>> > results
>> > with it.
>> >
>> > Thanks
>>
>>
date: Wed, 27 Feb 2008 16:02:29 -0800
author: Scott Schnoll [MSFT]
Re: Single SCR server to service multiple Exchange 2007 servers
Hi Scott,
We currently have Exchange 2003 implemented. We are moving to Exchange 2007
this summer on Windows Server 2008. We plan to have two CAS servers and two
HT servers. Those I can get them to spend money on having two each of
because they aren't that expensive. What I have mentioned previously is just
for the mailbox servers. I have done my best to educate on better solutions,
but in the end for the decision makers it comes down to cost. The mailbox
role is the most expensive to have with good redundancy since it contains all
the storage and high end servers.
I agree with everything you have stated and this has been preached more than
once to those that approve the budgets. I am in a position where I have to
leverage what I think I can do to get some form of redundancy at the absolute
minimum cost. Even then, I am not sure that will be accepted.
So I guess for my question rather than talk about what we "should" be doing
I have to keep it to "what I think I might be able to do". In this case
those are two very different things I know. With that said, can you offer
any general insight into using one SCR target server for multiple exchange
servers and around 10,000 mailboxes?
"Scott Schnoll [MSFT]" wrote:
> Thanks for the followup info and the kind words.
>
> What messaging system are you using now, and what sort of downtime have you
> experienced in the past? There's so much more that can go wrong than just a
> server being down. What about storage or other hardware failures? What
> about facilities failures (power, cooling, etc.)? What about the biggest
> cause of failures of all - human error?
>
> Sounds like from your post that you're the admin in charge of designing and
> implementing the solution. Part of your job (as you probably know) is to
> help educate the decision makers and set the expectations on the services
> provided by the solution (e.g., availability, uptime, etc., etc.). The core
> tenet of Exchange HA is to introduce redundancy throughout the environment.
> And it is the business needs that drive the design of the overall HA
> solution.
>
> In your organization, it sounds like you're saying the decision makers have
> decided that, although you have a lot of mailboxes, access to mail data or
> the ability to send/receive messages is not that critical for your users.
> If you're running Exchange now (e.g., Exchange 2003, etc.) then you may be
> able to get good usage statistics by collection a variety of performance
> counters. If you're not running Exchange, there still might be counters you
> can collect. Of course, if you have no messaging system now, then you might
> not know readily know what the specific availability needs are and will have
> to wait and see.
>
> BTW, one comment on your design of 3-4 servers and using CCR. When you
> deploy CCR, that will be two of your servers. That would leave you one or
> two servers for the Client Access and Hub Transport roles. If you only have
> one server for CAS/Hub, then you've got a bad single point of failure. If
> that server goes down you lose mail flow, and much, if not all, client
> access.
> --
> Regards,
>
> Scott Schnoll
> Microsoft Corporation
> This posting is provided "AS IS" with no warranties, and confers no
> rights. Please do not send email directly to this alias. This alias is for
> newsgroup purposes only.
>
>
> "Jon" wrote in message
> news:27025D8D-54EE-4BD4-8875-DA83295927C6@microsoft.com...
> >I understand where you are coming from but the organization I work for, as
> > large as they are, does not have much of that kind of information
> > available
> > and they don't seem overly concerned by the idea of an Exchange server
> > being
> > down for a while (although that story will change when the time comes I'm
> > afraid). As it sits now they do not want to spend any extra money on
> > redundancy or fault tolerance and this was my idea of something that I
> > could
> > attempt to sell them on where it would be the lowest cost to have some
> > type
> > of "poor man's clustering" in our environment and cover all our
> > servers/databases. I am not accustomed to being in an environment where
> > they
> > do not deem Exchange to be critical enough to put money into good
> > redundancy
> > configurations. It's been a struggle so far.
> >
> > I actually just watched your demos on the Exchange blog site about an hour
> > before I posted this so it was funny to see you reply to this. Great job
> > with that. The Exchange blog site is the best site ever for Exchange
> > admins.
> > Thanks for all the info that you all contribute there.
> >
> > "Scott Schnoll [MSFT]" wrote:
> >
> >> Let's first step back and focus on the business requirements. That
> >> really
> >> should be what drives the solution architecture and design.
> >>
> >> What sort of SLAs do you have for the messaging service and data
> >> availability?
> >>
> >> What are your RTO and RPO?
> >> --
> >> Regards,
> >>
> >> Scott Schnoll
> >> Microsoft Corporation
> >> This posting is provided "AS IS" with no warranties, and confers no
> >> rights. Please do not send email directly to this alias. This alias is
> >> for
> >> newsgroup purposes only.
> >>
> >>
> >> "Jon" wrote in message
> >> news:CFE8BCD1-FF48-445E-B948-2836E8AD5F61@microsoft.com...
> >> >I have an enviroment that is looking to have around 3 or 4 Exchange 2007
> >> > servers and 10,000 users. I would like to do CCR for all the servers
> >> > but
> >> > cannot convince the powers that be with the money to spend that much on
> >> > additional hardware and software. I was wondering if one bulky SCR
> >> > server
> >> > could handle 3 or 4 Exchange 2007 servers constantly pushing their logs
> >> > and
> >> > data to it. The idea is we can have this SCR server and use database
> >> > portability in the case of a disaster and this one box could support
> >> > being
> >> > the "clustered" server for all 3 or 4 other servers. I realize this is
> >> > not a
> >> > cluster and it would require manual steps to bring a database up if
> >> > necessary. I know every environment is different and a lot of this
> >> > depends
> >> > on how much mail flow we have and how large our databases are etc. I
> >> > just
> >> > wondered if using one SCR instance and sharing it for multiple live
> >> > Exchange
> >> > 2007 servers is something people have done and if they have had good
> >> > results
> >> > with it.
> >> >
> >> > Thanks
> >>
> >>
>
>
date: Thu, 28 Feb 2008 08:04:02 -0800
author: Jon
Re: Single SCR server to service multiple Exchange 2007 servers
Sure, very happy to help out any way I can here. But first, I did want to
suggest something (albeit something radical).
If your organization has not yet had an opportunity to measure their email
needs, you might see if you can get management to agree to a one-day turn
off of email. Just turn off the servers so no one can send/receive or see
any data they already have. This will help your organization determine how
necessary email is. Of course, don't just do this on your own - get
management buy in before you do anything like this. :-)
As for the design implications, the first thing you'll want to do is decide
how to configure the production environment in terms of number of servers,
sizing of those servers, and the storage needed to support your users. Have
you determined your users' disk IOPS needs yet? If not, you'll want to do
that first, as its critical for the design. You can predict your baseline
IOPS needs using two primary factors: the amount of database cache per user,
and the number of messages each user sends/receives each day. To estimate
the amount of database cache you'll have per user, subtract 2 GB from the
total amount of memory in the server (or that will be in the server), and
divide by number of users. Then, use the Exchange Profile Analyzer tool to
get user profile information. Once you have these two metrics, you can
enter them into the Exchange Server Storage Calculator and work out your
storage design. You will of course want to also design your storage
solution with sufficient capacity, as well.
Once you have the production side of things worked out, you need to work on
the recovery side of things. First, you need to determine what your plan
will be for activation of the SCR target databases. Of course, that will
depend largely on what failures you experience, and what model you're using
for your source and target (e.g., standalone or cluster). It works best if
the source and target are using the same model (e.g., source is CCR and
target is standby cluster; or source is standalone and target is
standalone). You may also need to pay attention to the version of Exchange
you use. You probably already know that you need the Enterprise Edition of
Exchange (and Windows) if you want to cluster Exchange. And if your SCR
target is a standby cluster you'll also need the Enterprise Edition of
Exchange (and Windows). If your SCR target is a standalone server, but you
need it to host more than 5 storage groups/databases, then you'll need the
Enterprise Edition of Exchange, as well.
SCR can be a little tricky when you consider its many-to-one, one-to-many,
and one-to-one options. When you're using the same SCR target for multiple
sources, you'll need to make sure that all systems are using the same
operating system (there is no cross-OS support in SCR), and that you have no
path collisions on the source and all of its targets. If the targets are
not hosting production databases, then its less tricky, but if there are
users hosted on the target, as well, you'll have more pathing to sort out.
This is some pretty general guidance, I know, but we've not really talked
too much in depth about the specifics of your design.
Let me know if you have any questions.
--
Regards,
Scott Schnoll
Microsoft Corporation
This posting is provided "AS IS" with no warranties, and confers no
rights. Please do not send email directly to this alias. This alias is for
newsgroup purposes only.
"Jon" wrote in message
news:348AD0B1-F84B-4A57-8572-CEBEAF3AF830@microsoft.com...
> Hi Scott,
>
> We currently have Exchange 2003 implemented. We are moving to Exchange
> 2007
> this summer on Windows Server 2008. We plan to have two CAS servers and
> two
> HT servers. Those I can get them to spend money on having two each of
> because they aren't that expensive. What I have mentioned previously is
> just
> for the mailbox servers. I have done my best to educate on better
> solutions,
> but in the end for the decision makers it comes down to cost. The mailbox
> role is the most expensive to have with good redundancy since it contains
> all
> the storage and high end servers.
>
> I agree with everything you have stated and this has been preached more
> than
> once to those that approve the budgets. I am in a position where I have
> to
> leverage what I think I can do to get some form of redundancy at the
> absolute
> minimum cost. Even then, I am not sure that will be accepted.
>
> So I guess for my question rather than talk about what we "should" be
> doing
> I have to keep it to "what I think I might be able to do". In this case
> those are two very different things I know. With that said, can you offer
> any general insight into using one SCR target server for multiple exchange
> servers and around 10,000 mailboxes?
>
> "Scott Schnoll [MSFT]" wrote:
>
>> Thanks for the followup info and the kind words.
>>
>> What messaging system are you using now, and what sort of downtime have
>> you
>> experienced in the past? There's so much more that can go wrong than
>> just a
>> server being down. What about storage or other hardware failures? What
>> about facilities failures (power, cooling, etc.)? What about the biggest
>> cause of failures of all - human error?
>>
>> Sounds like from your post that you're the admin in charge of designing
>> and
>> implementing the solution. Part of your job (as you probably know) is to
>> help educate the decision makers and set the expectations on the services
>> provided by the solution (e.g., availability, uptime, etc., etc.). The
>> core
>> tenet of Exchange HA is to introduce redundancy throughout the
>> environment.
>> And it is the business needs that drive the design of the overall HA
>> solution.
>>
>> In your organization, it sounds like you're saying the decision makers
>> have
>> decided that, although you have a lot of mailboxes, access to mail data
>> or
>> the ability to send/receive messages is not that critical for your users.
>> If you're running Exchange now (e.g., Exchange 2003, etc.) then you may
>> be
>> able to get good usage statistics by collection a variety of performance
>> counters. If you're not running Exchange, there still might be counters
>> you
>> can collect. Of course, if you have no messaging system now, then you
>> might
>> not know readily know what the specific availability needs are and will
>> have
>> to wait and see.
>>
>> BTW, one comment on your design of 3-4 servers and using CCR. When you
>> deploy CCR, that will be two of your servers. That would leave you one
>> or
>> two servers for the Client Access and Hub Transport roles. If you only
>> have
>> one server for CAS/Hub, then you've got a bad single point of failure.
>> If
>> that server goes down you lose mail flow, and much, if not all, client
>> access.
>> --
>> Regards,
>>
>> Scott Schnoll
>> Microsoft Corporation
>> This posting is provided "AS IS" with no warranties, and confers no
>> rights. Please do not send email directly to this alias. This alias is
>> for
>> newsgroup purposes only.
>>
>>
>> "Jon" wrote in message
>> news:27025D8D-54EE-4BD4-8875-DA83295927C6@microsoft.com...
>> >I understand where you are coming from but the organization I work for,
>> >as
>> > large as they are, does not have much of that kind of information
>> > available
>> > and they don't seem overly concerned by the idea of an Exchange server
>> > being
>> > down for a while (although that story will change when the time comes
>> > I'm
>> > afraid). As it sits now they do not want to spend any extra money on
>> > redundancy or fault tolerance and this was my idea of something that I
>> > could
>> > attempt to sell them on where it would be the lowest cost to have some
>> > type
>> > of "poor man's clustering" in our environment and cover all our
>> > servers/databases. I am not accustomed to being in an environment
>> > where
>> > they
>> > do not deem Exchange to be critical enough to put money into good
>> > redundancy
>> > configurations. It's been a struggle so far.
>> >
>> > I actually just watched your demos on the Exchange blog site about an
>> > hour
>> > before I posted this so it was funny to see you reply to this. Great
>> > job
>> > with that. The Exchange blog site is the best site ever for Exchange
>> > admins.
>> > Thanks for all the info that you all contribute there.
>> >
>> > "Scott Schnoll [MSFT]" wrote:
>> >
>> >> Let's first step back and focus on the business requirements. That
>> >> really
>> >> should be what drives the solution architecture and design.
>> >>
>> >> What sort of SLAs do you have for the messaging service and data
>> >> availability?
>> >>
>> >> What are your RTO and RPO?
>> >> --
>> >> Regards,
>> >>
>> >> Scott Schnoll
>> >> Microsoft Corporation
>> >> This posting is provided "AS IS" with no warranties, and confers no
>> >> rights. Please do not send email directly to this alias. This alias is
>> >> for
>> >> newsgroup purposes only.
>> >>
>> >>
>> >> "Jon" wrote in message
>> >> news:CFE8BCD1-FF48-445E-B948-2836E8AD5F61@microsoft.com...
>> >> >I have an enviroment that is looking to have around 3 or 4 Exchange
>> >> >2007
>> >> > servers and 10,000 users. I would like to do CCR for all the
>> >> > servers
>> >> > but
>> >> > cannot convince the powers that be with the money to spend that much
>> >> > on
>> >> > additional hardware and software. I was wondering if one bulky SCR
>> >> > server
>> >> > could handle 3 or 4 Exchange 2007 servers constantly pushing their
>> >> > logs
>> >> > and
>> >> > data to it. The idea is we can have this SCR server and use
>> >> > database
>> >> > portability in the case of a disaster and this one box could support
>> >> > being
>> >> > the "clustered" server for all 3 or 4 other servers. I realize this
>> >> > is
>> >> > not a
>> >> > cluster and it would require manual steps to bring a database up if
>> >> > necessary. I know every environment is different and a lot of this
>> >> > depends
>> >> > on how much mail flow we have and how large our databases are etc.
>> >> > I
>> >> > just
>> >> > wondered if using one SCR instance and sharing it for multiple live
>> >> > Exchange
>> >> > 2007 servers is something people have done and if they have had good
>> >> > results
>> >> > with it.
>> >> >
>> >> > Thanks
>> >>
>> >>
>>
>>
date: Thu, 28 Feb 2008 08:40:24 -0800
author: Scott Schnoll [MSFT]
Re: Single SCR server to service multiple Exchange 2007 servers
Thanks Scott, I will try to figure out what the IOPS will be using the data
you suggested I collect. We would be running about 4 or 5 Enterprise
Exchange servers and my idea was having an off-site machine be a standalone
SCR target for them. I would assume the thing I'd have to be most worried
about is that much information being written to disk on one server.
"Scott Schnoll [MSFT]" wrote:
> Sure, very happy to help out any way I can here. But first, I did want to
> suggest something (albeit something radical).
>
> If your organization has not yet had an opportunity to measure their email
> needs, you might see if you can get management to agree to a one-day turn
> off of email. Just turn off the servers so no one can send/receive or see
> any data they already have. This will help your organization determine how
> necessary email is. Of course, don't just do this on your own - get
> management buy in before you do anything like this. :-)
>
> As for the design implications, the first thing you'll want to do is decide
> how to configure the production environment in terms of number of servers,
> sizing of those servers, and the storage needed to support your users. Have
> you determined your users' disk IOPS needs yet? If not, you'll want to do
> that first, as its critical for the design. You can predict your baseline
> IOPS needs using two primary factors: the amount of database cache per user,
> and the number of messages each user sends/receives each day. To estimate
> the amount of database cache you'll have per user, subtract 2 GB from the
> total amount of memory in the server (or that will be in the server), and
> divide by number of users. Then, use the Exchange Profile Analyzer tool to
> get user profile information. Once you have these two metrics, you can
> enter them into the Exchange Server Storage Calculator and work out your
> storage design. You will of course want to also design your storage
> solution with sufficient capacity, as well.
>
> Once you have the production side of things worked out, you need to work on
> the recovery side of things. First, you need to determine what your plan
> will be for activation of the SCR target databases. Of course, that will
> depend largely on what failures you experience, and what model you're using
> for your source and target (e.g., standalone or cluster). It works best if
> the source and target are using the same model (e.g., source is CCR and
> target is standby cluster; or source is standalone and target is
> standalone). You may also need to pay attention to the version of Exchange
> you use. You probably already know that you need the Enterprise Edition of
> Exchange (and Windows) if you want to cluster Exchange. And if your SCR
> target is a standby cluster you'll also need the Enterprise Edition of
> Exchange (and Windows). If your SCR target is a standalone server, but you
> need it to host more than 5 storage groups/databases, then you'll need the
> Enterprise Edition of Exchange, as well.
>
> SCR can be a little tricky when you consider its many-to-one, one-to-many,
> and one-to-one options. When you're using the same SCR target for multiple
> sources, you'll need to make sure that all systems are using the same
> operating system (there is no cross-OS support in SCR), and that you have no
> path collisions on the source and all of its targets. If the targets are
> not hosting production databases, then its less tricky, but if there are
> users hosted on the target, as well, you'll have more pathing to sort out.
>
> This is some pretty general guidance, I know, but we've not really talked
> too much in depth about the specifics of your design.
>
> Let me know if you have any questions.
> --
> Regards,
>
> Scott Schnoll
> Microsoft Corporation
> This posting is provided "AS IS" with no warranties, and confers no
> rights. Please do not send email directly to this alias. This alias is for
> newsgroup purposes only.
>
>
> "Jon" wrote in message
> news:348AD0B1-F84B-4A57-8572-CEBEAF3AF830@microsoft.com...
> > Hi Scott,
> >
> > We currently have Exchange 2003 implemented. We are moving to Exchange
> > 2007
> > this summer on Windows Server 2008. We plan to have two CAS servers and
> > two
> > HT servers. Those I can get them to spend money on having two each of
> > because they aren't that expensive. What I have mentioned previously is
> > just
> > for the mailbox servers. I have done my best to educate on better
> > solutions,
> > but in the end for the decision makers it comes down to cost. The mailbox
> > role is the most expensive to have with good redundancy since it contains
> > all
> > the storage and high end servers.
> >
> > I agree with everything you have stated and this has been preached more
> > than
> > once to those that approve the budgets. I am in a position where I have
> > to
> > leverage what I think I can do to get some form of redundancy at the
> > absolute
> > minimum cost. Even then, I am not sure that will be accepted.
> >
> > So I guess for my question rather than talk about what we "should" be
> > doing
> > I have to keep it to "what I think I might be able to do". In this case
> > those are two very different things I know. With that said, can you offer
> > any general insight into using one SCR target server for multiple exchange
> > servers and around 10,000 mailboxes?
> >
> > "Scott Schnoll [MSFT]" wrote:
> >
> >> Thanks for the followup info and the kind words.
> >>
> >> What messaging system are you using now, and what sort of downtime have
> >> you
> >> experienced in the past? There's so much more that can go wrong than
> >> just a
> >> server being down. What about storage or other hardware failures? What
> >> about facilities failures (power, cooling, etc.)? What about the biggest
> >> cause of failures of all - human error?
> >>
> >> Sounds like from your post that you're the admin in charge of designing
> >> and
> >> implementing the solution. Part of your job (as you probably know) is to
> >> help educate the decision makers and set the expectations on the services
> >> provided by the solution (e.g., availability, uptime, etc., etc.). The
> >> core
> >> tenet of Exchange HA is to introduce redundancy throughout the
> >> environment.
> >> And it is the business needs that drive the design of the overall HA
> >> solution.
> >>
> >> In your organization, it sounds like you're saying the decision makers
> >> have
> >> decided that, although you have a lot of mailboxes, access to mail data
> >> or
> >> the ability to send/receive messages is not that critical for your users.
> >> If you're running Exchange now (e.g., Exchange 2003, etc.) then you may
> >> be
> >> able to get good usage statistics by collection a variety of performance
> >> counters. If you're not running Exchange, there still might be counters
> >> you
> >> can collect. Of course, if you have no messaging system now, then you
> >> might
> >> not know readily know what the specific availability needs are and will
> >> have
> >> to wait and see.
> >>
> >> BTW, one comment on your design of 3-4 servers and using CCR. When you
> >> deploy CCR, that will be two of your servers. That would leave you one
> >> or
> >> two servers for the Client Access and Hub Transport roles. If you only
> >> have
> >> one server for CAS/Hub, then you've got a bad single point of failure.
> >> If
> >> that server goes down you lose mail flow, and much, if not all, client
> >> access.
> >> --
> >> Regards,
> >>
> >> Scott Schnoll
> >> Microsoft Corporation
> >> This posting is provided "AS IS" with no warranties, and confers no
> >> rights. Please do not send email directly to this alias. This alias is
> >> for
> >> newsgroup purposes only.
> >>
> >>
> >> "Jon" wrote in message
> >> news:27025D8D-54EE-4BD4-8875-DA83295927C6@microsoft.com...
> >> >I understand where you are coming from but the organization I work for,
> >> >as
> >> > large as they are, does not have much of that kind of information
> >> > available
> >> > and they don't seem overly concerned by the idea of an Exchange server
> >> > being
> >> > down for a while (although that story will change when the time comes
> >> > I'm
> >> > afraid). As it sits now they do not want to spend any extra money on
> >> > redundancy or fault tolerance and this was my idea of something that I
> >> > could
> >> > attempt to sell them on where it would be the lowest cost to have some
> >> > type
> >> > of "poor man's clustering" in our environment and cover all our
> >> > servers/databases. I am not accustomed to being in an environment
> >> > where
> >> > they
> >> > do not deem Exchange to be critical enough to put money into good
> >> > redundancy
> >> > configurations. It's been a struggle so far.
> >> >
> >> > I actually just watched your demos on the Exchange blog site about an
> >> > hour
> >> > before I posted this so it was funny to see you reply to this. Great
> >> > job
> >> > with that. The Exchange blog site is the best site ever for Exchange
> >> > admins.
> >> > Thanks for all the info that you all contribute there.
> >> >
> >> > "Scott Schnoll [MSFT]" wrote:
> >> >
> >> >> Let's first step back and focus on the business requirements. That
> >> >> really
> >> >> should be what drives the solution architecture and design.
> >> >>
> >> >> What sort of SLAs do you have for the messaging service and data
> >> >> availability?
> >> >>
> >> >> What are your RTO and RPO?
> >> >> --
> >> >> Regards,
> >> >>
> >> >> Scott Schnoll
> >> >> Microsoft Corporation
> >> >> This posting is provided "AS IS" with no warranties, and confers no
> >> >> rights. Please do not send email directly to this alias. This alias is
> >> >> for
> >> >> newsgroup purposes only.
> >> >>
> >> >>
> >> >> "Jon" wrote in message
> >> >> news:CFE8BCD1-FF48-445E-B948-2836E8AD5F61@microsoft.com...
> >> >> >I have an enviroment that is looking to have around 3 or 4 Exchange
> >> >> >2007
> >> >> > servers and 10,000 users. I would like to do CCR for all the
> >> >> > servers
> >> >> > but
> >> >> > cannot convince the powers that be with the money to spend that much
> >> >> > on
> >> >> > additional hardware and software. I was wondering if one bulky SCR
> >> >> > server
> >> >> > could handle 3 or 4 Exchange 2007 servers constantly pushing their
> >> >> > logs
> >> >> > and
> >> >> > data to it. The idea is we can have this SCR server and use
> >> >> > database
> >> >> > portability in the case of a disaster and this one box could support
> >> >> > being
> >> >> > the "clustered" server for all 3 or 4 other servers. I realize this
> >> >> > is
> >> >> > not a
> >> >> > cluster and it would require manual steps to bring a database up if
> >> >> > necessary. I know every environment is different and a lot of this
> >> >> > depends
> >> >> > on how much mail flow we have and how large our databases are etc.
> >> >> > I
> >> >> > just
> >> >> > wondered if using one SCR instance and sharing it for multiple live
> >> >> > Exchange
> >> >> > 2007 servers is something people have done and if they have had good
> >> >> > results
> >> >> > with it.
> >> >> >
> >> >> > Thanks
> >> >>
> >> >>
> >>
> >>
>
>
date: Wed, 5 Mar 2008 11:34:01 -0800
author: Jon
|
|
