CAS array without load balancing, which IP to point the CAS array to?

OK, if i have (2) Exchange 2010 SP1 boxes and both have CAS, HT and Mailbox roles and I've configured a DAG between the 2 and I've created a CAS array and both boxes are members. I want to know what is my best option for which IP address I should point the CAS array to?  Remember, I don't have NLB or a hardware load balancer at this point (although I hope to procure/deploy a Kemp appliance in the future).

Right now the CAS array is named Outlook and I have an internal DNS record for Outlook that points to 1 of the 2010 boxes.  I've set the TTL value to 5 minutes for this record.  If I had a failure of this CAS I figured I could manually change the IP to that of the 2nd 2010 box.  I've read that I could assign a 2nd DNS record for the CAS array and setup DNS Round Robin, but that could result in half of the users getting pointed to a failed device in the event of a failure - is this a better approach that what I currently have?  Seems like half getting pointed to a failed device is better than all getting pointed to a failed device?  If there was a failure, what should I do - just remove the DNS entry for the failed server or change it so they both point to the same box?

Or, I've seen a lot of recommendations to point the CAS array to the IP address of the DAG, this is where my uncertainty comes into play, what exactly is this achieving?  Would this allow the clients to always be pointing to the live server in the event of a failure?  Would this automatically send clients directly to the box that hosts their live mailbox (vs the other server that would host their passive mailbox)?

I've seen so many comments about all three of these options, but I haven't seen a detailed description and comparison of them all.  And I'm looking for a clear understanding of what pointing the CAS array to the DAG IP will do.

Thanks.
wpstechAsked:
Who is Participating?
I wear a lot of hats...

"The solutions and answers provided on Experts Exchange have been extremely helpful to me over the last few years. I wear a lot of hats - Developer, Database Administrator, Help Desk, etc., so I know a lot of things but not a lot about one thing. Experts Exchange gives me answers from people who do know a lot about one thing, in a easy to use platform." -Todd S.

Jian An LimSolutions ArchitectCommented:
http://blogs.kraftkennedy.com/index.php/2009/11/25/configuring-nlb-for-exchange-2010-cas-load-balancing/

i wonder can you create one ..

but bear in mind, DAG is not really good to sit with CAS and HT.

really hard to play with :P


0
seb_ackerCommented:
You cant use nlb on your infrastructure.
One way is to use DNS round robin.

But that isn't efficient. Usually, i let it as you have it, and try to convince the financial staff that you need a hardware load balancer
0
RadweldCommented:
Bascially you may as well just leave the CASARRAY pointing to your DAG member hosting activation preference 1 databases (the main one) if the server dies, you can just update the CASARRAY (DNS Alias) to pointo the second server. Ideally you dont want to use NLB, you should be looking at Hardware Load balancers as NLB isnt service aware and NLB will still direct traffic to a failed node.
0
Ultimate Tool Kit for Technology Solution Provider

Broken down into practical pointers and step-by-step instructions, the IT Service Excellence Tool Kit delivers expert advice for technology solution providers. Get your free copy now.

wpstechAuthor Commented:
radweld - i was planning to have approximately 6 databases and have 3 live/active and 3 passive on each of the 2 servers.  if I do this do you still recommend that i keep it as is until we can acquire a hardware load balancer like the kemp technologies appliance?  In your recommendation you recommended pointing the IP to the DAG member that is hosting the active DB, but I will have active DBs on both members.

also, i'm still not seeing recommendations/opinions on pointing the CAS array to the DAG IP (virtual IP assigned to the DAG when it is created)...
0
Adam BrownSr Solutions ArchitectCommented:
The IP address assigned in DNS for the CAS Array FQDN should be set to point to one of the CAS servers with a very short TTL in DNS so you can change the IP address on the fly if one of the servers fails. Once you get a Load Balancer, you would change that IP to point to the Load Balancer. I can see pointing at the DAG VIP causing problems, but I've never actually done that so I can't tell you for certain if it would or wouldn't work, but pointing it to one of the servers' real IP addresses will work for certain.

I'm a little curious about your reasons for having 6 active/passive pairs, though. With an infrastructure that doesn't have the roles separated out onto different servers that many databases might not serve much of a purpose. How big are you planning to let these databases get?

Also, I say this because there are some special considerations to be made when you have all the server roles installed on your two HA servers. You may want to take a look at the following for information: http://technet.microsoft.com/en-us/library/ff699049.aspx 
0
wpstechAuthor Commented:
acbrown2010 - please offer more detail for your curiosity for what I described.  I was just thinking along the lines that separating the mailboxes into logical groups of databases would make sense and reduce downtime if a database failure/corruption ever occurred.  Am I going down the wrong path here?  I only had 2 servers to work with and they were exactly identical, so I just assumed that the best use of the hardware was to install all 3 roles on both servers.   I felt like leaving all of the live databases on 1 server and only using the other server for passive copies would be kind of a waste of that hardware...

I'm open to any/all opinions/recommendations/advice, so please feel free to lend it.  I havent migrated mailboxes yet so I'm not 100% set in stone with this topology.

thanks.
0
Adam BrownSr Solutions ArchitectCommented:
The benefit you'll see from having multiple databases depends on a few things.
1. What backup method are you using? If you are using a backup system that is capable of single item restores in Exchange, the benefit of having smaller sized databases for quicker restores kind of disappears. If you're using hard drive based storage for your backups, there is also not as much benefit to having smaller databases as if you were to have a tape backup or something similar. If you can budget in a high-quality Exchange Backup system, you can save yourself a lot of trouble.
2. With a DAG, you're much less likely to run into a situation where both copies of a database fails, so having multiple databases running at the same time doesn't have as much benefit for keeping some users up in the event of a failure. If the DB goes down, it goes up on the passive server.
3. The size limit on databases that MS recommends is 2TB. But if you have that much email, you really need to expand outward a bit.

Having multiple databases can help, but you need to weigh the advantages against the increased administrative overhead of managing users on those databases. Whether it makes sense or not depends on the number of users you have on your network.

Another option you may want to look into is Virtualization. If you have two servers, and since you're going to have to use Windows Server 2008 Enterprise, you have the option of running each server as a virtual host. Enterprise licenses allow you to run up to 4 VMs on a single license, so you can have up to 4 VMs on each server. This would allow you to separate out the roles if you want to. It should be noted, though, that you would need to get an exchange server license for each VM if you use that method, so it'll be more expensive for licensing. But that is just an option and whether you go that route depends heavily on how many users you have on your network. It does, however, open up the option of having a third virtualized Mailbox server as a lagged database copy that you can use for reducing the impact of database corruption.
0

Experts Exchange Solution brought to you by

Your issues matter to us.

Facing a tech roadblock? Get the help and guidance you need from experienced professionals who care. Ask your question anytime, anywhere, with no hassle.

Start your 7-day free trial
RadweldCommented:
Virtualising could be an option, that way you can leverage virtual load balancers or seperate cas/hub servers running NLB. NLB  would be cheaper but so is the experience. A load balancer is by far the best. The va16 offering from load balancers.org works out at £1500 for the pair. If you choose NLB then you need a pair of standard windows licenses (included with your host) and a pair of exchange standard licenses as you will be running four exchange servers.

Databases in exchange 2010 are supported upto 18tb but the recommendation when in a dag is 2tb. When not in a dag the recommendation is 250gb. You might only need one database and one copy on the other server.
0
wpstechAuthor Commented:
on the current Ex2003 environment there are 4 separate mailstores, and they are essentially setup this way for administrative purposes (size quota differences, message limits, etc).  We are a school system with about 800 users, but our overall database size is small (less than 100GB) because we limit the size of the mailboxes.  The current mailstores were setup for managing them by role (administrative personnel, elementary teachers, middle school teachers, high school teachers, etc).  This will begin to change as we move to 2010 and allow for larger mailboxes.  

I gave Hyper-V a slight thought before building the 2 servers as they are now, but decided against it.  I just decided against virtualizing the production Exchange system, although perhaps it would have been just fine.  

Ignoring the new size allowances from a DB perspective with 2010, and considering the scenario of 800 users whose average mailbox will likely remain less than 500MB, that's still less than 500GB total.  With that said, do you think that having a total of 6 live databases in a 2 server DAG (3 live/3 passive on each server) is still not a good option? Performance wise where does this hurt me?

thank you so much for all of your input/advice btw
0
It's more than this solution.Get answers and train to solve all your tech problems - anytime, anywhere.Try it for free Edge Out The Competitionfor your dream job with proven skills and certifications.Get started today Stand Outas the employee with proven skills.Start learning today for free Move Your Career Forwardwith certification training in the latest technologies.Start your trial today
Exchange

From novice to tech pro — start learning today.

Question has a verified solution.

Are you are experiencing a similar issue? Get a personalized answer when you ask a related question.

Have a better answer? Share it in a comment.