Link to home
Start Free TrialLog in
Avatar of mrjking2000
mrjking2000

asked on

Setting up unique Hyper-V enviornment

Good Morning Everyone,

I have a slightly unique deployment opportunity that I'm seeking some input on.  I have a client that for 2 months out of the year is a large public attraction, corn maze, pumpkin patch, zombie paintball hunting and ghost town etc...

Down time during this 8 week season is not an option.

I have migrated them to Server 2012 R2 and have done a slight bit of the pre-work for how I'd like to see this setup.  Their primary server is strong enough to run Hyper-V and at least a single virtual server.  So I have the physical machine in a standalone workgroup with the only thing being ran off that box right now one single DC in a virtual deployment.  that DC is of course 2012 R2 as well, and it also hosts quickbooks point of sale, a time clock program for employee check-in/out and also just quickbooks pro database manager.

it's a pretty small footprint, and over the last 48 hours has been running great.

I want to introduce some high-availability into this environment where the virtual will live migrate to another physical machine in a different building, but since it is a farm and there is not any existing fire suppression, I don't want all the components in the same building.  Right now two buildings are the primary and secondary network hubs (per say) as one camera server resides in a dedicated locked room with a 24 port netgear managed switch running gigabit and the other Hyper-V host server sits in the office also hooked up to a 24 port netgear gigabit managed switch and this is also where the router is for the internet.

nowhere on the network currently is there any storage attached, either NAS or SAN.

If we deployed some kind of attached storage to the network, it would sit in one location or the other which in the event of a fire, kills the hyper-V as the storage would be gone that the VHD runs off of.

How can I (or is it possible) to setup a storage location on each server, without using a single point such as a NAS, so in the event of a network drop or catastrophic failure such as fire that the virtual would move over to the other server?

My familiarity with hyper-v is usually just spinning up virtuals on a single very powerful server.  adding in the storage I understand for migration, but that's all usually in the same room.  How do I get what I'd like to see with this shared across two physical buildings with no single point of failure?

We do have 3 cat6 feeds between each building in which only 1 is being used.  So I have flexibility in "off-lan" private linking if needed.

Thanks!
Avatar of David Johnson, CD
David Johnson, CD
Flag of Canada image

Your scenario is well suited to Azure Scaling pay for what you use architecture. For the two months using express route would be fairly inexpensive.

you should have the following
san1      <-> san 2
    |     X           |
server1  <->     server 2

San1 replicates with San 2, server 1 replicates with server 2,
network connection from server 1 to san 1 and san 2 and server 2
Storage should be using a separate switch than the servers
Avatar of mrjking2000
mrjking2000

ASKER

I should follow up and mention that offsite isn't a good idea in this particular environment.  even with SLA'd bonded T1 service they are not reliable and have gone down.  Quickbooks POS allows transactions to run offline and once the internet comes back, they process.

what kind of on-site deployment could this tailor to.

Thanks a ton!
Do you understand the simple diagram that I posted?
Suggest you take this free course https://mva.microsoft.com/en-US/training-courses/failover-clustering-in-windows-server-2012-r2-8489
ASKER CERTIFIED SOLUTION
Avatar of dipersp
dipersp
Flag of United States of America image

Link to home
membership
This solution is only available to members.
To access this solution, you must be a member of Experts Exchange.
Start Free Trial
I'd think that 5 minutes would be acceptable.  I will check into both suggestions.
We use Hyper-V replica extensively for moving data offsite for DR purposes (No different than doing it over the Internet/WAN/LAN in your situation.)  Works great and has tested out very well in our DR testing.

If you have any questions on it, ask away.  I'd personally start with Hyper-V replica, since it's included and free and works well.
First of all, if the host OS is Standard then 2 VMs can be set up and be properly licensed. I suggest putting the LoB and its backend on a second VM. Leave the DC to being a DC.

I have an EE article that may help to answer some questions: Some Hyper-V Hardware and Software Best Practices.

On recoverability, Hyper-V Replica is great for static related workloads running on the VMs. Exchange and SQL for example want nothing to do with Replica given the lag between source and replica. I suspect that this would be the case for their LoB that is active in nature.

So, what do we do?

There are two or maybe three realistic options:

1: A very good backup setup that is _tested_ on a quarterly basis. Tested meaning bare-metal or hypervisor restored to production ready. This is how we have our standalone hosts set up via SuperSpeed USB3 external drive for the backups.

2: A Clustered Storage Spaces setup. The indicated cluster in that post is running at a 15 seat accounting firm. The cost of downtime during personal tax season and the multiple year-end corporate tax seasons up here can be _very_ costly. Cost for this cluster is quite reasonable.

3: Run the setup in a fully redundant public cloud. Two Internet connections (hard wired and cell) could provide the necessary Internet connectivity and redundancy.
Hi all,

was "offline" over the weekend so I am going to really start looking at this hard soon.  appreciate the feedback.

thanks,
After a few meetings with the client, we have decided that the best way to go in this particular situation now is the built in replication to a 2nd machine in a 2nd building onsite.  Thanks for all the feedback everyone.