smcauley
asked on
Exchange 2010 Toplogy
Hi Experts. I would like some advise on our current exchange 2010 topology and the best fit for us moving forward.
We are a relatively small company support 150 users, mostly in one location. We have a main site and a disaster recovery site and are currently running the following:-
Main Site:-
2 x physical CAS servers
2 x virtual DAG servers
DR Site:-
1 x virtual CAS server
2 x virtual DAG servers
We would like if possible to collapse our roles and consolidate some of these servers along with virtualizng our CAS servers. My understanding is that to support load balancing on the CAS servers in a virtual environment we would need a physical hardware load balancer?
Ideally we would like to probably run 2 servers here each having all roles and one in our DR site. Is this configuration supported or a rational approach?
The other variable is we run Veeam Backup & Replication and can replicate our Exchange across to the DR site and failover there if necessary. Do we need to run another server in DR if we have that failover option?
Thanks in advance!
We are a relatively small company support 150 users, mostly in one location. We have a main site and a disaster recovery site and are currently running the following:-
Main Site:-
2 x physical CAS servers
2 x virtual DAG servers
DR Site:-
1 x virtual CAS server
2 x virtual DAG servers
We would like if possible to collapse our roles and consolidate some of these servers along with virtualizng our CAS servers. My understanding is that to support load balancing on the CAS servers in a virtual environment we would need a physical hardware load balancer?
Ideally we would like to probably run 2 servers here each having all roles and one in our DR site. Is this configuration supported or a rational approach?
The other variable is we run Veeam Backup & Replication and can replicate our Exchange across to the DR site and failover there if necessary. Do we need to run another server in DR if we have that failover option?
Thanks in advance!
ash007
As you are using DAG so it will give you very High Avalability but also for CAS load balancing you can use windows Network load balancer that will work fine no need for hardware load balancer.
ASKER
Thanks Ash,
I thought I read somewhere that what you suggested wasn't supported by Microsoft in virtual environments?
I thought I read somewhere that what you suggested wasn't supported by Microsoft in virtual environments?
Hello,
Could you please give more details of :
- your virtulization provider (vmware, hyperv, xen,...)
- how many DAG do you plan :
1 stretched DAG (with 4 members) across your HQ and PRA sites
or 2 DAGs (with 2 members on each site)
Some configuration of NLB and virtualization are supported, but usually the support of NLB relies more on the virtualisation provider than Microsoft. For example, vmware supports NLB in "Multicast mode" only, for Exchange CAS
Could you please give more details of :
- your virtulization provider (vmware, hyperv, xen,...)
- how many DAG do you plan :
1 stretched DAG (with 4 members) across your HQ and PRA sites
or 2 DAGs (with 2 members on each site)
Some configuration of NLB and virtualization are supported, but usually the support of NLB relies more on the virtualisation provider than Microsoft. For example, vmware supports NLB in "Multicast mode" only, for Exchange CAS
I am running a (4) Exchange 2010 DAG server configuartion behind a Hardware based load balancer (Citrix NetScaler) divided between two sites (DC/DR). One of the three servers is configured as a LAG copy. All servers are virtualized (VMware 4.x - upgrading to 5.x soon) in a boot from SAN configuration using RDMs for their data drives. The File Share Witness is located at the primary site where the bulk of the users reside. All Exchange roles are collapsed and equal on each server. We have about 250 users and about 350 mailboxes.
Fyi... A three server DAG configuration won't automatically failover to the DR site. This is why we choose the four server configuration. Also, I prefer a hardware Load Balancer that is independent of your virtual infrastructure. Also, according to Microsoft this is considered a backup-less Exchange configuration therefore you wouldn't require backups using Veeam or any other backup software. However, you can use it as an extra measure in case of Database corruptions. In our case, we are going to leverage CommVault for email backup/recovery, archive, and e-Discovery. At the SAN level the backup data will be replicated. This is pretty much overkill, but we will be doing it for extra measure.
Fyi... A three server DAG configuration won't automatically failover to the DR site. This is why we choose the four server configuration. Also, I prefer a hardware Load Balancer that is independent of your virtual infrastructure. Also, according to Microsoft this is considered a backup-less Exchange configuration therefore you wouldn't require backups using Veeam or any other backup software. However, you can use it as an extra measure in case of Database corruptions. In our case, we are going to leverage CommVault for email backup/recovery, archive, and e-Discovery. At the SAN level the backup data will be replicated. This is pretty much overkill, but we will be doing it for extra measure.
Additional notes, we are moving from a 2 cluster VMware environment (DC/DR) to a 4 ESX Cluster server split between sites with 1 Exchange 2010 DAG in each ESX cluster. This configuration will be used from an HA perspective to spread out the risk. This is required for our environment since our primary clusters are using HP C7000 chassis w/ BL490s (can be a single point of failure) and secondary clusters will be independent HP DL380 servers.
ASKER
Thanks for the comments guys. To clarify our setup we are running on the following:-
Main Site:-
3 x HP DL380 G7 ESXi hosts running vSphere 5 Enterprise Plus (HA, DRS enabled)
2 x StorageWorks 8/8 FC switches
1 x HP MSA 2000 G3 FC
DR Site:-
1 x HP DL380 G5 running vSphere 4 Standard (soon to be upgraded)
1 x HP MSA 2312i iscsi
Exchange config is 1 x stretched DAG with 4 members between the sites. FSW is at the primary site.
@gsmartin. Setup sounds close to what we have or are looking at albeit you are supporting a lot more users. I just wonder if what we are doing is overkill for our size. Perhaps not though and our best option is a hardware load balancer with 4 Exchange Servers (2 per side) with fully collapsed roles?
Main Site:-
3 x HP DL380 G7 ESXi hosts running vSphere 5 Enterprise Plus (HA, DRS enabled)
2 x StorageWorks 8/8 FC switches
1 x HP MSA 2000 G3 FC
DR Site:-
1 x HP DL380 G5 running vSphere 4 Standard (soon to be upgraded)
1 x HP MSA 2312i iscsi
Exchange config is 1 x stretched DAG with 4 members between the sites. FSW is at the primary site.
@gsmartin. Setup sounds close to what we have or are looking at albeit you are supporting a lot more users. I just wonder if what we are doing is overkill for our size. Perhaps not though and our best option is a hardware load balancer with 4 Exchange Servers (2 per side) with fully collapsed roles?
Typically, you'd say it's even overkill for my company size, but what's more important to the Business is to ensure mail is always available. Email these days is one of the most critical services for businesses. Therefore, given it's critical nature you either host it internally with a DAG configuration or outsource it to Microsoft Online Hosted services; which we are only using for Anti-virus and Anti-Spam filtering. This is much better than hosting it internally because you can't match the same level of protection hosted internally.
How many information stors (exchange databases) are you using? We have ours broken down into 4 databases and are thinking of dividing it even further. We had experience a couple database corruptions that we succesfully recovered from, but are planning thin the databases out further to minimize the impact in case of future events. The main databases are divided between the two primary servers at our DC.
How many information stors (exchange databases) are you using? We have ours broken down into 4 databases and are thinking of dividing it even further. We had experience a couple database corruptions that we succesfully recovered from, but are planning thin the databases out further to minimize the impact in case of future events. The main databases are divided between the two primary servers at our DC.
ASKER
We are running 3 DBS in total, one for standard users, 1 for execs and an online archive for all users. Email is the most critical component of our business so it needs to be HA 24/7. Just hoped we could do it with less servers and still have that redundancy! Maybe the best we can hope for is the collapsed CAS roles with a H/W load balancer and 4 servers.
Agreed. All four of my servers have equal roles CAS, HUB, Mailboxes, UM (will add role soon), and EDGE. My original configuration was 2 CAS/HUB/EDGE and 3 MAILBOX servers divided accross both sites until I learned about the DAG configuration. I first considered the 3 server DAG, but didn't like the manual failover aspect.
FYI... One caveat to having your FSW at the your Primary site is if the site goes down the failover site (DR) won't be able to go online. Personally, I am looking at implementing a DFSR solution to this issue.
FYI... One caveat to having your FSW at the your Primary site is if the site goes down the failover site (DR) won't be able to go online. Personally, I am looking at implementing a DFSR solution to this issue.
ASKER
Yes, we do use replication as the moment to copy the FSW to the DR site.
One other question in regards to what you said about backups not being necessary in your current setup (although I know you do). How would logs ever get truncated without a backup?
Thanks.
One other question in regards to what you said about backups not being necessary in your current setup (although I know you do). How would logs ever get truncated without a backup?
Thanks.
ASKER CERTIFIED SOLUTION
membership
This solution is only available to members.
To access this solution, you must be a member of Experts Exchange.
ASKER
Thanks for all your comments gsmartin. We will collapse the CAS roles which will drop 3 servers for us and go with the hardware load balancer with 4 Exchange DAG servers.
Interesting stuff about the circular logging, not sure I'm brave enough to recommend that path though!
Interesting stuff about the circular logging, not sure I'm brave enough to recommend that path though!
I wasn't sure about circular logging as well, but our PS vendor also recommended it.
One direction I am changing since my last post is with Archiving and E-Discovery (CommVault term); I have decided to leverage Exchange 2010 capabilities vs CommVault this will cut back on our CommVault licensing, and lose de-dup capabilities. However, we be to leverage de-dup in the near future with our SAN. We will most likely only use CommVault for backup and recovery as a fail-safe; still deciding this. This should be okay assuming our SAN and CommVault can play harmonious with each-other, which they having a good partnership and I expect it won't be an issue given previous conversations with our vendor.
Good luck with your project!
One direction I am changing since my last post is with Archiving and E-Discovery (CommVault term); I have decided to leverage Exchange 2010 capabilities vs CommVault this will cut back on our CommVault licensing, and lose de-dup capabilities. However, we be to leverage de-dup in the near future with our SAN. We will most likely only use CommVault for backup and recovery as a fail-safe; still deciding this. This should be okay assuming our SAN and CommVault can play harmonious with each-other, which they having a good partnership and I expect it won't be an issue given previous conversations with our vendor.
Good luck with your project!