Exch2010 DAG - Bandwidth / connection type

We're trying to revisit our DAG setup that is active and passive nodes in two separate state lcoations.  GA and NY.  

Both active and passive nodes hold the Hub, Client, and Mailbox roles of the five roles.
Each site connect via hw vpn established over 10 m fiber.
175 mailboxes.
4 Dbs(50,55,60,47 gb) with log files ranging from 2,3,5, and 10g presently/at this second.

questions:
1. DAG works best over MPLS or VPN hw like Sonicwall or WatchGard or checkpoint? If can do/invest in MPLS should we just default to implementing that.

*when something happens at our passive node site getting the CQL right and db back over to GA is grueling and once we even had to have things sneakermailed bc it just wouldn't seed and/or come back(copy) on line in GA, successully.

2. Do we have the roles necessary for users to actually get their mail from the passive node(NY) during a failover, because as things are (inherited) mail barely works when this happens. OWA doesn't and mail may or may trickle into some users. mostly not.

Thanks
dee30Asked:
Who is Participating?
 
GiladnCommented:
I hope I understand, correct me if i'm wrong  about topology:

1. DAG works best over MPLS, you get much better performace.

2. you should diagnose this problem, might be even DNS related, do you clients rely somehow on active node with static config? can you click crtl+right click over the outlook icon   (when passive is active). Connection Status should appear. Are any of the connections going straight to the passive node?

3. what happens when you reboot the active node, do clients get outlook client reconnect automaticly? did you check routing and verified matrics?

Hope this helps,

G
0
 
dee30Author Commented:
starting to think about possibly introducing some colo hosted solution just for the DAG.  something in-state where the passive node is over metroE or something.  We wouldn't require the MPLS services, would continue maintaing the hw tunnels between our locs and just have a colo hosted site house just the passive DAG member(#2).  Thoughts

Also, is the DAG doable over a hw vpn tunnel we establish though if we were just sticking to that question?
0
 
GiladnCommented:
it will work, depanding on bandwidth upstream/downstream and amount of users.
you may encounter very slow access at some times..

G
0
 
Simon Butler (Sembee)ConsultantCommented:
Most of the DAG implementations I have done are over a VPN of some description. The idea being that if the link fails we can get something in place (anything) to reconnect.
Hopefully you have an RPC CAS array in place.

If you do put something in a colo, then one option is to have all traffic (including OWA, Outlook Anywhere etc) come in via that server. Then in the event of a failover, everything is going to the live server. If it was just the internet connection that had failed you could de-queue the email over a 3g card!

Simon.
0
 
dee30Author Commented:
just divied up points... not real conclusion to this... working through it with some inputs from above is all doing at this point.  thank you.
0
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.

All Courses

From novice to tech pro — start learning today.