High Availability Server 2012 R2

GECs36711 used Ask the Experts™
Hi All,

I would just like to find out what it would take to create a High Availability Domain in 2012 R2. Doing so with the Minimum about of hardware and software where possible.

Please structure it around the following if needed and any other service that is not mentioned:
-Servers, Hyper V, clustering, failovers, no single point of failure, DFS, CSV Shares, Multi path, Virtual Disks, Storage pools, Replica's VMM. Lets for now say without System Centre Components.

Please indicate and briefly explain why used; From Start to Finish.

Make a Highly Available Domain, DC, FMSO roles and shares with the minimums.
(E.g. Two Servers, one at each site, one site goes down everything is still available)
Simplicity is key.
I'm having trouble understanding where everything connect together.

Much Appreciated for any of the help.
Watch Question

Do more with

Expert Office
EXPERT OFFICE® is a registered trademark of EXPERTS EXCHANGE®
kevinhsiehNetwork Engineer

Depends on how highly available you want to make things. Highly available also isn't the same as fault tolerant.

Domain services such as AD, DHCP, and DNS are generally already HA when you configure them properly. Of all of those, systems tend to be less tolerant of their primary DNS server being unavailable.

For file shares, DFS replication combined with DFS namespace might fit the bill, however be aware of file locking issues, and you can experience data loss for files that have not replicated.

HA for clustering involves some sort of shared storage, be it physical shared storage like quality SAN/NAS/SAS DAS, or virtual shared storage such as Nutanix, Starwind Software, HP StorVirtual VSA, etc. Using CSV is just part of clustering.

Every server needs to be connected to two separate switches. There should be redundant routers for every segment. If going over a WAN there should be two separate circuits, preferably from two different carriers, even better if they are not both in the street, or at least not the same street.

For high availability everything should be on two separate circuits, preferably two different UPS, or at least one UPS and 1 utility power. There should also be 1, preferably more than 1 generators on site.

Many applications such as SQL, Exchange, AD have their own HA capabilities built in for replication, log shipping, etc. Those are always preferred if they are available (ie can you afford the licensing and other infrastructure costs).

You have asked a very broad question. Is there something more specific that you are looking for?
Philip ElderTechnical Architect - HA/Compute/Storage

Simple way is two virtualization hosts that has one DC each.


Sounds good, would you kindly give me more detail and specifics.

Maybe if you could please specify more on Windows Server 2012 R2 Roles, Features and Services. Basically the backend to a simple Highly Available DC (FMSO roles, DHCP, DNS, AD) and best route for creating shares.(Let say between two sites that are identical in structure). Please no single points of failure as mentioned. I know it could be a vast answer but i'd appreciate just then the outlines. eg. Shares; add File Server services, add iSCSI target and initiator(Why), add multi path(Why), Configure.....
Technical Architect - HA/Compute/Storage
You could have two identically configured servers set up with 2012 R2 and the Hyper-V Role.

Set up a VM on each server that will host ADDS, DNS, DHCP in failover mode, and if needed file and print services (we tend to put these on another VM).

From there, you could set up VMs for other tasks such as Exchange, SQL, LoBs (Line-of-Business apps), SQL, and others.

A business critical VM hosting files and folders could be replicated between the two hosts to keep the business online if the hardware fails on one host.
as noted by some of the experts above, this is a very broad question.

in general though, I suspect you're over complicating it a little.
AD facilities are high-availability by design, so the AD,DNS,DFS facilties can be achived by simply having 2 DCs running those services.
Exchange & SQL do too,

I therefore feel that clustering, shared storage etc is overkill for what you may need. if you'd like to consider it please let us know and we can give some info on options.

I recommend setting up a good spec standalone hyper-v host and creating several VMs on it, and having a 2nd similar hyper-v host elsewhere with additional VMs on.
You can easily get DC, DNS, DFS replication between VMs on each server as that's what it's designed for. Exchange & SQL are a little more complicated but still straightforward to run across two VMs (consider DAGS for mailbox resilience)

All you're then left with is any applications out of your control. You can setup VM replication between the two hyper-v systems to take a copy of the VM that you can fire up in the event of failure.

Do more with

Expert Office
Submit tech questions to Ask the Experts™ at any time to receive solutions, advice, and new ideas from leading industry professionals.

Start 7-Day Free Trial