none
DAG with active passive database configuration

    Question

  • We are planning to deploy four exchange 2010 servers each one will have (mailbox+Cas+hub transport) roles.

    2 of these servers will be in the primary data-center and the other 2 will be in the DR data-center.

    I am thinking one DAG with active/passive site resilient. one active copy of the database on each primary server and 1 passive copy on each DR server.

    The question is will this configuration work or do i need to have multiple copies of the database on each server? 

    if i do need multiple copies then will the passive copy take the same amount of space as the active copy so if i have a passive and active copy on the same server will i need twice the space ?

    i am asking this because i don't have a lot of storage on these servers and i may need to add more storage  if i need to have multiple copies on each server.

     

    Thanks,

     

    Ben 

     

     

    Wednesday, September 29, 2010 11:36 PM

Answers

  • How dos that sound ?

     

    i have a DS3 45 MB connection to my DR, i will have to see how much bandwidth is being used by other traffic.

    anyone have a guess on how much  of bandwidth will 125 mailbox server requires for replication ?

     

    You need to start using this; none of us here can truly answer those questions as they require a lot more environmental data. :) What may work in one organization's environment with a T-3 may not work in another one depending on what else is going on already inside that pipe.

     


    Microsoft Premier Field Engineer, Exchange
    MCSA 2000/2003, CCNA
    MCITP: Enterprise Messaging Administrator 2010
    Former Microsoft MVP, Exchange Server
    My posts are provided “AS IS” with no guarantees, no warranties, and they confer no rights.
    Sunday, October 03, 2010 2:40 PM

All replies

  • Ben,

    I would recommend you read this, http://technet.microsoft.com/en-us/library/dd638137.aspx  has good information about planning for site resiliency.

    So if I understand you will have

    Site A - Two Exchange Servers with MB,HUB, CAS

    Site B - Two Exchange Servers with MB,HUb,CAS

    With a DAG and your configuration you could have up to a a total of 4 copies per DB, one active and the other 3 passive. 

    How many DBs are you planning to deploy?

    What are your goals?  High Availability for the PRimary site and Disaster Recovery to the DR site?  or just Disaster Recovery to the DR site?

     


    Sean McNeill Microsoft Gold Partner http://staterainfrastructure.blogspot.com/
    Thursday, September 30, 2010 12:24 AM
  •  

    Thank you Sean,

    After reading the article you suggested it is all starting to make sense. 2 databases 4 copies per DB , one active and the other 3 passive. 

    Thursday, September 30, 2010 10:26 PM
  • Just to be exact, you can only ever have one database copy that is active at any one point in time. All the other copies are passive and are candidates to become active if they can be maintained in a healthy condition. The three passive copies that you are considering are in this category.

    A server can only host a single copy of a database. Therefore, as I understand what you're proposing, you will have the active copy and one passive copy in Site A and two passive copies in Site B (the DR site). 

    - Tony

    Thursday, September 30, 2010 10:47 PM
  • You could also look at having a lag copy over to your DR site, which could give you up to 2 weeks of rollback ability, in case a log somewhere along the way gets corrupted.  Having your 2 Mailbox servers at your main site replicate between each other gives you pretty decent availability, and the lag copy would give you DR ability.  If you're going to have live databases replicating to your DR site, you need to make sure you have really good bandwidth between the sites.
    Thursday, September 30, 2010 10:49 PM
  • Ben,

    In addition, you will want to read up on CAS/DAG which cannot be shared on the same set of servers. If you decide later to make your CAS servers redundant (creating a cas array), you won't be able to given your proposed setup. You cannot place a CAS array on the same servers in which you have a DAG configuration. Just a design consideration you may want to think about.

    -David

    Thursday, September 30, 2010 11:51 PM
  • Thanks having a lag copy is a good idea. so i am changing my design to have 2 active databases and each database will have two passive copies one at second server at site A and one at third server at site B and a lag copy at fourth server at site B. 

    How dos that sound ?

     

    i have a DS3 45 MB connection to my DR, i will have to see how much bandwidth is being used by other traffic.

    anyone have a guess on how much  of bandwidth will 125 mailbox server requires for replication ?

    Friday, October 01, 2010 5:55 AM
  • Are you sure about that David ? correct me if  i am wrong but from what i read you can have a CAS array even if you have CAS/DAG on the same server the only caveat is that you must use a hardware load balancer to load balance your CAS array. 
    Friday, October 01, 2010 6:01 AM
  • That is correct. A hardware load balancer in front of the servers can hold CAS/DAG on the same set.

    I should have been more specific and said "if you plan on using wnlb" for CAS, which I assumed you were doing since Microsoft states to use a hardware load balancer if you have more than eight client access servers in a single active directory site. You have only four, which is I why I ____ u me d ;-).

     

    Friday, October 01, 2010 5:39 PM
  • Ben,

    Replication traffic is more due to mail traffic than anything, so if your environment have high mail traffic then that will affect repliction traffic.  For a DAG to function accross a WAN you need less than 250 MS of latency per Microsoft, real world would recommend no more than 150 MS latency.  Sounds like you will be fine with a 45MB DS3 connecting the sites.

    My experiance is to forgo LAG copies and allow for your DR site to be updated immediately.  With the new Log replication and replay into DBs, my experiance as have implemented over 10 Exchange 2010 environements is that DB corruption from Log replication is non-existant.  In the off chance a DB does corrupt then go to tape or other backup to restore.  MY recommendation is to keep the copies as close to in-sync as possible in the event you must failover to the DR site.

     

    Hope this helps and by no means is this meant to be gospel for your environment, many variables and decisions to be made on how you replicate your DAG and plan for recovery and failover.


    Sean McNeill Microsoft Gold Partner http://staterainfrastructure.blogspot.com/
    Saturday, October 02, 2010 3:48 AM
  • How dos that sound ?

     

    i have a DS3 45 MB connection to my DR, i will have to see how much bandwidth is being used by other traffic.

    anyone have a guess on how much  of bandwidth will 125 mailbox server requires for replication ?

     

    You need to start using this; none of us here can truly answer those questions as they require a lot more environmental data. :) What may work in one organization's environment with a T-3 may not work in another one depending on what else is going on already inside that pipe.

     


    Microsoft Premier Field Engineer, Exchange
    MCSA 2000/2003, CCNA
    MCITP: Enterprise Messaging Administrator 2010
    Former Microsoft MVP, Exchange Server
    My posts are provided “AS IS” with no guarantees, no warranties, and they confer no rights.
    Sunday, October 03, 2010 2:40 PM
  • With the new Log replication and replay into DBs, my experiance as have implemented over 10 Exchange 2010 environements is that DB corruption from Log replication is non-existant.  In the off chance a DB does corrupt then go to tape or other backup to restore.  MY recommendation is to keep the copies as close to in-sync as possible in the event you must failover to the DR site.

     

    There are two types of corruption to understand; Physical and Logical. 

    • Physical: When the database structure itself from an ESE perspective is no longer valid.
    • Logical: When the data within the database is no longer valid, but the ESE structures are valid. So the physical view of the data is messed up.

    Physical corruption cannot replicate. This is what the Exchange replication system protects us against. It also helps save us from this with bad block detection and the ability to reseed a single database page from a remote DB copy.

    Logical corruption can replicate. Perhaps a client did something whacky to an item it shouldn't have, or someone modified an item incorrectly with MFCMapi. Since the database structure itself is good, Exchange will happily replicate these changes. This is the type of corruption lag copies can potentially help with, but the difficult thing is knowing *when* it happened. Then you have to get to the lag copy before it replays the logs from that date/time. I don't know many people who can actually identify that actual point in time.

    Sooooo to borrow some words from RSIV on the product team there are various ways to deal with logical corruption.

    • Move the mailbox, thereby discarding bad items.
    • Use single item recovery to restore an original item if it was a result of editing an item.
    • Use the Calendar Repair Assistant to detect/repair corrupted meeting items.
    • Use New-MailboxRepairRequest to fix search folders, item counts, parent/child folder issues.
    • Keep a point in time (PIT) backup and if all of the above doesn't work, restore the PIT backup of the DB to a RDB.

    Microsoft Premier Field Engineer, Exchange
    MCSA 2000/2003, CCNA
    MCITP: Enterprise Messaging Administrator 2010
    Former Microsoft MVP, Exchange Server
    My posts are provided “AS IS” with no guarantees, no warranties, and they confer no rights.
    Sunday, October 03, 2010 2:56 PM