DR Stragegies for Medium sized business
Morning Experts,
I'm currently working on a project where we will be migrating our ERP system to a new set of servers in the coming months, and I'm trying to come up with a plan for disaster recovery.
The high level requirements are as follows.
1. SQL Server 2008 R2 (Virtual)
2. Application Server (Virtual)
3. Load Balanced Terminal Servers (Physical)
4. Off-site hyperV host connected to the main Datacenter via 6mbps MPLS.
I'm not too concerned about the terminal servers, but trying to come up with some ideas to protect the SQL and APplication server.
I have setup a test environment with two Server 2008 machines running HyperV for testing.
1. I tested replicating the VHD files between the servers using DPM 2010. While this works, the recovery process is kind of complicated.
2. Currently testing a Hyper-V cluster using SteelEye Datakeeper to replicate the VM's between servers. So far the recovery seems to be far easier than DPM.
3. Next Step is to evaluate Double Take Availability.
Anybody have any words of wisdom or pitfalls I should watch out for?
I'm currently working on a project where we will be migrating our ERP system to a new set of servers in the coming months, and I'm trying to come up with a plan for disaster recovery.
The high level requirements are as follows.
1. SQL Server 2008 R2 (Virtual)
2. Application Server (Virtual)
3. Load Balanced Terminal Servers (Physical)
4. Off-site hyperV host connected to the main Datacenter via 6mbps MPLS.
I'm not too concerned about the terminal servers, but trying to come up with some ideas to protect the SQL and APplication server.
I have setup a test environment with two Server 2008 machines running HyperV for testing.
1. I tested replicating the VHD files between the servers using DPM 2010. While this works, the recovery process is kind of complicated.
2. Currently testing a Hyper-V cluster using SteelEye Datakeeper to replicate the VM's between servers. So far the recovery seems to be far easier than DPM.
3. Next Step is to evaluate Double Take Availability.
Anybody have any words of wisdom or pitfalls I should watch out for?
ASKER CERTIFIED SOLUTION
membership
This solution is only available to members.
To access this solution, you must be a member of Experts Exchange.
Okay, so have you actually considered using the SANs, to replicate and mirror the LUNs?
You DR site will then always have an uptodate copy of the LUNs, ready to mount for DR purposes.
SAN to SAN replication will be done in the background, without any other third party replicators.
(or Snapshots and Virtual Disk Copy I believe is what Dell call it!)
You DR site will then always have an uptodate copy of the LUNs, ready to mount for DR purposes.
SAN to SAN replication will be done in the background, without any other third party replicators.
(or Snapshots and Virtual Disk Copy I believe is what Dell call it!)
ASKER
hanccocka,
I'm not sure if the MD3000i and the 3200i are compatible with each other as far as Replication is concerned, but yes we have considered it.
Also, for what it is worth, my Dell rep told me the "snapshot" capability of the 3200i is not application consistent so there is a possibility of data corruption.
These SAN's weren't specifically specc'ed out with virtualization and DR in mind, they were more for low cost & high performance storage.
What the Dell rep mentioned is correct, they are crash consistent. If that is a concern, but we've never seen any data corruption. But it's nice for him to highlight the risks!
ASKER
If I had a bottomless budget for the project I would probably buy two Equallogics PS4000X 's and replicate all the data that way.
I don't think the budget will allow for that, especially not for a small VM environment (Figure 2 - 5 VM's)
I should have mentioned that our Terminal Server(s) support up to 120 concurrent users. I know we can probably squeeze them all into one, but I want to have a second physical box for redundancy.
SOLUTION
membership
This solution is only available to members.
To access this solution, you must be a member of Experts Exchange.
ASKER
kevin,
A SQL mirror and standby servers was my first thought as well. The problem with that is the app server has a number of files that change on a fairly regular basis. The majority of the business data resides in SQL though.
A SQL mirror and standby servers was my first thought as well. The problem with that is the app server has a number of files that change on a fairly regular basis. The majority of the business data resides in SQL though.
I would look at SteeleEye for the application servers and see how that looks, and then consider it for SQL and RD servers. If SteeleEye looks good, I wouldn't bother evaluating DoubleTake unless you find it to be a bit cheaper than SteeleEye or from the literature it does something that the SteeleEye doesn't that you want.
ASKER
The biggest issue I can find so far with SteelEye is each VM requires it's own volume.
ASKER