Virtualized Exchange on Amazon EC2 for a DR site - has anyone tried this ?

Posted on 2011-03-09
Medium Priority
Last Modified: 2012-08-10
Hello Experts.

We are exploring some options for a DR site for an Exchange 2010 installation.
This is part of Exchange 2010 migration plan.

Present - Exchange 2003

Proposed Setup
Virtualized Exchange 2010
on 2 Dell r710 rack mount servers. Backend Dell iSCSI SAN
MBX and CAS virtualized
Using vmware vSphere

AD's on physical server, not virutalized.

DR Site
Amazon EC2 running the same instance Exchange 2010 instance

Does it make sense to have a DAG Replication going on with a VM instance in EC2 ?

Please feel free to suggest options / alternatives with High Availability / Price considerations.

I am not sure if 500 points will be sufficient for your thoughts :)

Thanks for your input.


Question by:sunnyc7
  • 6
  • 3
LVL 28

Author Comment

ID: 35088650
I think this is closest to what I am planning to do.

LVL 32

Accepted Solution

endital1097 earned 2000 total points
ID: 35090091
What is the network bandwidth between the primary site and Amazon? The maximum round trip latency for a DAG is 250ms.
What is the average number of logs generated daily and during peak times?
How are clients going to connect? Can the network handle the number of RPC connections or will everyone be configured to use Outlook Anywhere?
How will you handle a primary site failure for Internet access to mail (if available to users)?
Is one of the two dell servers for the DC? If so, the design is attempting high availability with a single CAS?

Could you add a load balancer (F5) and use the two dell servers to virtualize everything? One DC, one MBX, one CAS/HUB VM on each node. You could have up to a four node DAG in the primary site with two copies of each database. Then if you still wanted, you could add Amazon to the equation. This would give you redundancy within the site for scheduled maintenance. The load balancer would also give you redundancy for your CAS servers. Of course you would also need additional storage for this design.

LVL 28

Author Comment

ID: 35113531
Thanks a lot for dropping in on this. I truly appreciate.
Between posting this question and now, I had another similar case come-up and I have been on call with multiple vendors / customers. Sorry for the delay in replying here.

Some other pieces of info:
Really small Network -  40 users.
Bandwidth Full T1.
> Need to test the roundtrip latency between the site and EC2-N.Va. TODO

Logs generated
I have numbers for Ex2003, as I havent started yet on the migration to 2010 / co-existence
I have less than 30 logs per day. Normal. Daily backups - never missed a backup. All logs flushing properly.

Can the network handle the number of RPC connections or will everyone be configured to use Outlook Anywhere?
>> Less than 5 users for Anywhere

How will you handle a primary site failure for Internet access to mail (if available to users)?
>> Very important question and a real good one.

This is what I have in mind -- feel free to correct me / make recommendations.
Primary WAN - Dual WANs hit a Cisco 29XX.
2 VLAN's - one to Amazon EC2 / Second one to the PIX 506.
If VLAN1 fails - failover to EC2/Doubletake. <-- working on this design with a Cisco guy. i have a call scheduled.

The amount of users doesnt necessitate F5.
I will most probably go for WNLB and load balance the CAS (Not necessary I think given the number of users..)

Is one of the two dell servers for the DC? If so, the design is attempting high availability with a single CAS?
>> DC's.
I have 3 other Dell PE 2900 - 1 is a physical DC.
I was planning to go with vmWare Essentials Plus which supports upto 6 hosts, and allows for vmotion FT and DRS

Node1 - Windows 2008 DC - with GC Role
Node2 - EX1-HT-CAS-MBX
Node3 - EX2-MBX
Node4- App.ServerX
Node5 - nothing as of now
Node6 - nothing as of now.

VM Images of Node1 and Node3

2 MBX - Node2/Node3
placing DC-GC Windows 2008 in the VM itself.
will set-up WNLB between node2 and node3.
I can make Node3 - CAS/MBX or HT/CAS/MBX  - but I dont see the point for 40 users :(

Move I/O intensive to Physical machine's
Anti-spam running from a dell PE - physical
Backup Running from a dell PE - physical

What do you think ?
Making Bulk Changes to Active Directory

Watch this video to see how easy it is to make mass changes to Active Directory from an external text file without using complicated scripts.

LVL 32

Assisted Solution

endital1097 earned 2000 total points
ID: 35114044
That is a really low log count so I don't think log generation will be an isssue
Is this design overkill for such a small user base? Unless you are experiencing a lot of unplanned outages out of your control which last an extended amount of time and email must be highly available.

Why put so much effort into making the databases highly available if your access point (CAS) is not redundant?
DAG is not supported on VM cluster
You also need to make sure multicast is used within VM for WNLB

EC2 has VM images of node1 and 3? What exactly does this mean? I thought you were going to have a cross site DAG?

I'll think about this some more and post additional comments later (need to get kids to bed)
LVL 28

Author Comment

ID: 35114081
I meant Amazon EC2 in the cloud will host a VM copy of DC and a copy of DAG

I was thinking that @ design overkill.
LVL 28

Author Comment

ID: 35114085
*database copy

No lagged copies FYI.
LVL 32

Assisted Solution

endital1097 earned 2000 total points
ID: 35121580
Overall this will take some effort to put into place, but there is no reason why it shouldn't work
I still don't see a response on the CAS server design
*assumed no lagged copies based on the topology

Once again based on the user count I think it is a very expensive solution
Do they expect an ROI with this project or are they failing to meet an SLA that is costly?
LVL 28

Author Comment

ID: 35126744
Its more sla focussed right now than roi - I would say.
I want to give an offsite dr-site as well as no dr and 2 cas-mbx.

I think 1 cas should be sufficient given the number of users but would deploy 2.
Have you used ADC's like citrix netscaler vpx / coyotepoint ?
LVL 28

Author Closing Comment

ID: 35137803
I am going to close this question. Talking to some private cloud providers on importing VM's directly and checking what networking options they provide.

Thanks Jim

Featured Post

Simplify Active Directory Administration

Administration of Active Directory does not have to be hard.  Too often what should be a simple task is made more difficult than it needs to be.The solution?  Hyena from SystemTools Software.  With ease-of-use as well as powerful importing and bulk updating capabilities.

Question has a verified solution.

If you are experiencing a similar issue, please ask a related question

Microsoft Jet database engine errors can crop up out of nowhere to disrupt the working of the Exchange server. Decoding why a particular error occurs goes a long way in determining the right solution for it.
There are literally thousands of Exchange recovery applications out there. So how do you end up picking one that’s ideal for your business & purpose? By carefully scouting the product’s features, the benefits it offers you, & reading ample reviews f…
The video tutorial explains the basics of the Exchange server Database Availability groups. The components of this video include: 1. Automatic Failover 2. Failover Clustering 3. Active Manager
Exchange organizations may use the Journaling Agent of the Transport Service to archive messages going through Exchange. However, if the Transport Service is integrated with some email content management application (such as an antispam), the admini…
Suggested Courses

579 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