Solved

Exchange2010 to 2013 upgrade  and HA question

Posted on 2015-01-07
30
47 Views
Last Modified: 2016-07-05
I am planning to upgrade my Exchange2010SP3 with DAG of 3 members.
Right now I have 2 plans below are the details. servers having 64GB memory

Design 1
production site
2013  Mailbox servers (2 nos)
2013  CAS servers  (2 nos) with WNLB
DR site
2013 Mailbox/CAS (single hardware with all roles)

Design 2
Production site
2013 Mailbox/CAS (single hardware all roles)  with hardware load balancer
2013 Mailbox/CAS (single hardware all roles)
DR site
2013 Mailbox/CAS (single hardware all roles)

Please suggest which design is more efficient and reliable.
If both not good please suggest another design.


What is the easiest and recommended way to move mailboxes from 2010 to 2013 ?

Your guidance is appreciated.
0
Comment
Question by:-MAS
  • 11
  • 5
  • 4
  • +5
30 Comments
 
LVL 13

Expert Comment

by:Andy M
ID: 40535138
Personally I'd go with the second option - Exchange 2013 works fine with all roles installed on the same system (I've setup a few exch 2013 systems with DAGS and never had to split out the roles to separate servers).

Note that for full automated failover ideally you need the witness server at a third site that has access to both production and dr.

Moving the mailboxes is easy - setup the Exchange 2013 servers to run alongside the Exchange 2010 ones then just use the mailbox move command to move the mailbox to another database.
0
 
LVL 9

Expert Comment

by:Marshal Hubs
ID: 40535246
I'm totally agree with Andy, even I'd go with the second option. You can move the mailboxes by using powershell scripts or if you are not comfortable in executing powershell cmdlets, then you can try Stellar Mailbox Extractor For Exchange Server, through which you can directly export mailbox from one Exchange Server to another Exchange Server by just selecting the mailbox which you want to migrate and then by entering the details of the Exchange Server 2013. Rest software will do by own quickly.
0
 
LVL 24

Author Comment

by:-MAS
ID: 40535267
Can you justify/explain the reason of taking the second option?

Bcz I have to submit to the management.

Really appreciate if you can write pros and cons
0
 
LVL 13

Accepted Solution

by:
Andy M earned 250 total points
ID: 40535615
The second option gives more resilience in the event of a server issue as you effectively have two servers at the production site (which will also help with load balancing).

In the first option if the CAS server goes down your users will need to connect to the DR site to get email (which depending on situation/configuration) will connect back to the production site (mailbox server) to access the databases.
In the second option if one server goes down the users will just switch to the second server (though effectively your hardware load balancer would cover this) - they should only need to connect to the DR if the entire production site is down (or in the event both production exchange servers goes down which would be very unfortunate).

Also please note that WNLB does have some restrictions for use on Exchange 2013: (taken from Microsoft article - http://technet.microsoft.com/en-us/library/jj898588%28v=exchg.150%29.aspx):

* WNLB can't be used on Exchange servers where mailbox DAGs are also being used because WNLB is incompatible with Windows failover clustering. If you're using an Exchange 2013 DAG and you want to use WNLB, you need to have the Client Access server role and the Mailbox server role running on separate servers.

 * WNLB doesn't detect service outages. WNLB only detects server outages by IP address. This means that if a particular web service, such as Outlook Web App, fails, but the server is still functioning, WNLB won’t detect the failure and will still route requests to that Client Access server. Manual intervention is required to remove the Client Access server experiencing the outage from the load balancing pool.

 * Using WNLB can result in port flooding, which can overwhelm networks.

 * Because WNLB only performs client affinity using the source IP address, it's not an effective solution when the source IP pool is small. This can occur when the source IP pool is from a remote network subnet or when your organization is using network address translation.
0
 
LVL 24

Author Comment

by:-MAS
ID: 40535650
Thanks Andy for your guidance.

WNLB only used if the CAS in seperate hardwares (as in Design1)

You mean you selected design2 only because of load balancer?
FYI we already have a load balancer in place.
0
 
LVL 53

Expert Comment

by:Will Szymkowski
ID: 40535782
Just to add to what the others have already mentioned is that Exchange 2013 is designed to run more efficently with all roles on the same server. This reduces cost for server licensing etc. Also as we all know WNLB is not a good option for load balancing due to all of times limitations.

It is also stated that Microsoft does not recommend WNLB for Production environments and only lab environments.

Will.
0
 
LVL 13

Expert Comment

by:Andy M
ID: 40537533
No, just noting about WNLB, which also Will has noted is not ideal for production environments as well.

Option 2 effectively gives you 3 fully working Exchange servers (2 production and one dr) so greater resiliance and reduces load over VPN/WAN connections in the event of a failed server, where as option 1 only gives you 2 (one production consisting of 2 servers - one cas server and one mailbox server, and one dr).

Alternatively to save costs (if you are needing to purchase all the hardware/licenses/etc) you could have one server at production with all roles on and one server at the dr with all roles on.

Also as clarified by Will, 2013 is designed to run effectively with all roles on the same server.
0
 
LVL 24

Author Comment

by:-MAS
ID: 40537597
Thanks to all.
Apart from WNLB what are the other Pros and cons on the below design.

Design 1
production site
2013  Mailbox servers (2 nos)
2013  CAS servers  (2 nos) with HLBalancer
DR site
2013 Mailbox/CAS (single hardware with all roles)

Design 2
Production site
2013 Mailbox/CAS (single hardware all roles)  with hardware load balancer
2013 Mailbox/CAS (single hardware all roles)
DR site
2013 Mailbox/CAS (single hardware all roles)

Please suggest which design is more efficient and reliable.
If both not good please suggest another design.

Please advice how we manage the DR server traffic.
i.e. Do I have to add as a member in HLB or other way to manage the traffic?
Automatic failover or manual fail over is better ?
Do I have to configure dynamic quorum to have automatic fail over if link between production and DR supports?
0
 
LVL 53

Assisted Solution

by:Will Szymkowski
Will Szymkowski earned 250 total points
ID: 40543109
I think all of the appropriate points have been made here already.
- WNLB not supported for production environments
- Exchange 2013 is designed to run more efficently on same hardware (all roles)
- You have high availability with option 2
- Hardware load balancers can do layer 4 and 7 which provides grandular monitoring of specfic Exchange services.

Will.
0
 
LVL 24

Author Comment

by:-MAS
ID: 40543148
Thanks will and  Amit
I am not clear on how we manage the traffic between DR site and production site.
0
 
LVL 42

Expert Comment

by:kevinhsieh
ID: 40543174
Your option 1 requires 5 copies of Windows and Exchange, option 2 only requires 3.

I don't know what you mean by manage the traffic between the two sites. I personally use QoS to make sure that the DAG seeding and replication doesn't swamp my network connection. What are you referring to?
0
 
LVL 24

Author Comment

by:-MAS
ID: 40543181
I mean do I need to add as a member in HLB?
If yes is there any preference to activate as the DR server as the final server to activate?

I need DR site to be activated only when both the CAS servers goes down, how we achieve this?
0
 
LVL 42

Expert Comment

by:kevinhsieh
ID: 40543207
Don't use NLB. Use your load balancer.

There are activation preferences on each database for each DAG member. The system knows how to fail over from one to the other automatically.
0
How your wiki can always stay up-to-date

Quip doubles as a “living” wiki and a project management tool that evolves with your organization. As you finish projects in Quip, the work remains, easily accessible to all team members, new and old.
- Increase transparency
- Onboard new hires faster
- Access from mobile/offline

 
LVL 53

Expert Comment

by:Will Szymkowski
ID: 40543210
I think you are talking about 2 different things here. When you are working with a DAG it is highly recommended that you have a Production Network for mailbox access and a completely isolated network for replication traffic for the DAG. The Hardware Load Balancer is used only for your CAS side. Depending on how you want to set it up Layer 4 or Layer 7 you will need to create your URL's to point to your hardware load balancer VIP and you would then assign your CAS servers to this VIP as a pool member.

Can't really go into details becase different vendors have different method of setting up load balanceing based on the hardware, but what i have listed above is a high level of what needs to be done.

I would personally take the time (if you hardware load balancer provides this) is setting up Layer 7 for Exchange. It does take more time and configuration but you can provide better monitoring for specific services then doing it using Layer 4.

Layer 4 will for Exchange 2013 just fine but it just all depends on how you want to configure the load balancing of your CAS servers.

Will.
0
 
LVL 24

Author Comment

by:-MAS
ID: 40543223
My concern is only CAS server load balancing not mailbox .
Mailbx will be replicated
0
 
LVL 53

Expert Comment

by:Will Szymkowski
ID: 40543236
Just make sure that Mailbox replication traffic is not on the production network or you will see delays on your entire network when traffic is being replicated.

As for the CAS setup I think all of your questions have been answered. HLB is the way to go for production purposes. Specific setup of HLB is all based on the vendor you use.

Will.
0
 
LVL 24

Author Comment

by:-MAS
ID: 40543680
One more question please
What about shadow redundancy?
Each copy of all email will go to the DR server as well or is there any workaround to avoid this?
Or it is normal ?
0
 
LVL 42

Expert Comment

by:kevinhsieh
ID: 40543691
You control on a database level which servers the database is replicated to. Is there any reason why you wouldn't replicate all of your databases to the DR server?
0
 
LVL 24

Author Comment

by:-MAS
ID: 40543697
Thanks for your quick reply
I am talking about shadow redundancy feature of Exchange
0
 
LVL 42

Expert Comment

by:kevinhsieh
ID: 40545961
From my reading about shadow redundancy, if the email is received by a server in your primary location, a second copy will be stored by the other server in the same site until the message is delivered into the mailbox. If a message is received by the DR server, then a copy of that message should also get copied to a server in your primary site so that there are two copies as it flows through your system. This generally shouldn't be an issue.

How much mail do you expect to be storing in total, and how much bandwidth do you have between systems? I have around 4 TB of mail, and many of the databases are between 200 and 300 GB. It can take quite a while to resync a database over the WAN (or even locally). To avoid having to reseed over the WAN I literally once drive a SAN and server from one site to another, reseeded all of my databases, and then drove it all back because otherwise it was going to make me a week or two to get things resynced. WAN optimization such as Riverbed or Silverpeak can help. So does a fast connection. :-)
0
 
LVL 24

Author Comment

by:-MAS
ID: 40546126
Many thanks for your reply.
So you mean it is ok we keep shadow redundancy?

Anyway I am reading few articles regarding the bandwidth requirement for DAG. I will get back to u

Thanks a lot kevinhsieh
0
 
LVL 35

Expert Comment

by:Mahesh
ID: 40547356
Beginning from Exchange 2010 SP1 I guess MS Exchange product team start recommending multirole servers to save hardware, to get maximum benefits from hardware etc and to increase high availability
http://windowsitpro.com/blog/why-installing-multirole-exchange-2013-server-best-option
0
 
LVL 41

Expert Comment

by:Amit
ID: 40547410
Hi MAS, site resiliency is a vast topic. If you are looking for serious implementation, I suggest you to hire a consultant, who know how to do it and what all required.
0
 
LVL 42

Expert Comment

by:kevinhsieh
ID: 40548109
I wouldn't turn off shadow redundancy unless you really know what you are doing and why. I wouldn't try to outsmart the system.
0
 
LVL 18

Expert Comment

by:irweazelwallis
ID: 40549220
you can do what we have done which is to reduce the shadow redundancy timeout

http://technet.microsoft.com/en-gb/library/dd351046(v=exchg.141).aspx

you can reduce it down 8 hours to cut down the performance impact
0
 
LVL 24

Author Comment

by:-MAS
ID: 40550579
@irweazelwallis
Many thanks
I am not clear on your statement "you can reduce it down 8 hours to cut down the performance impact"

What is the default value and what is the advantage if we change(whether it is decrease  or increase)?
0
 
LVL 24

Author Closing Comment

by:-MAS
ID: 41691929
Thanks
0

Featured Post

Free book by J.Peter Bruzzese, Microsoft MVP

Are you using Office 365? Trying to set up email signatures but you’re struggling with transport rules and connectors? Let renowned Microsoft MVP J.Peter Bruzzese show you how in this exclusive e-book on Office 365 email signatures. Better yet, it’s free!

Join & Write a Comment

Resolve DNS query failed errors for Exchange
OfficeMate Freezes on login or does not load after login credentials are input.
In this video we show how to create an email address policy in Exchange 2013. We show this process by using the Exchange Admin Center. Log into Exchange Admin Center.:  First we need to log into the Exchange Admin Center. Navigate to the Mail Flow…
In this Micro Tutorial viewers will learn how they can get their files copied out from their unbootable system without need to use recovery services. As an example non-bootable Windows 2012R2 installation is used which has boot problems.

760 members asked questions and received personalized solutions in the past 7 days.

Join the community of 500,000 technology professionals and ask your questions.

Join & Ask a Question

Need Help in Real-Time?

Connect with top rated Experts

20 Experts available now in Live!

Get 1:1 Help Now