Avatar of Paul
Paul asked on

Windows Server DR using replication / Failover accross WAN

Firstly, forgive the ridiculousness of this question.

Scenario is this

A server at 1 location. Operating a virtual Host OS (VMware / Hyper-V) with 2 or 3 guests

1. Windows Server (2012?) with Standard Roles such as DC, File server, Exchange Email, SQL Server roles)
2. Windows Terminal Server (2012?)
3. Possible Legacy virtual system operating Server 2003 with legacy database application

He wants the above to replicate over the internet to an identical setup at another site for DR purposes. Internet connection at primary site is 10/10 and secondary site is 50/1 The amount of file store data involved in total excluding OS files is around 100GB. Daily changes will be below 1GB.

Im looking for advice on whats the best way to achieve this. Preferably using internal storage. I am fairly certain the answer is 3rd party software which enhances the capabilities of cluster failover, or negates it altogether. Automatic and instantaneous failover would be preferable, but small downtime (5-30m) should be OK but again, without intervention. Even better if that software also handles backups, and if not, an idea of how such a setup would interact with inbuilt Windows Server Backup.

For the record, no hardware or software has been purchased yet, so theres a blank slate from that perspective. You have carte blanche in terms of suggesting the appropriate setup balancing reliability and functionality against cost.

Thanks for your input people,

Kind Regards.
Disaster RecoveryWindows Server 2012VMware

Avatar of undefined
Last Comment

8/22/2022 - Mon
Gerald Connolly

So Internet connections are 10MbitPS /10MbitPS and 50MbitPS/1MbitPS?

So with 10 bits per byte over a comms link that is approx 1MBPS/1MBPS and 5MBPS/100KBPS

The 100KBytes/sec is going to be a bit of bottleneck!

How do you plan to sync the two sites, 100GB+ over a 1MB/s link is going to take a very long time. Once it is synched, 1GB per day shouldnt stretch the link too much, but is there going to be other traffic on this link as well?

So failover shouldnt be too much of a truma, but what about fail-back? have you worked out how long that is going to take?

Hi Gerald

Thanks for coming. Yes, those figures are correct. The 100KBPS is going to be a serious pain, but as I say my opinion isnt really being asked for. I am being engaged to deploy a specific setup as per somebody elses design. IMO, if the site does fail over for any extended period of time we might even just go in and restore a backup of the DR server onto the main server once its repaired.

Oh yeh, and did i mention they only have 15 users?

Anyways, whats the best method to do DR replication over a wan link? I am under the impression that all built in MS technologies pretty much wont work without unnecessary and costly SAN/iSCSI ETC
Brad Bouchard

There are a couple of ways I would do this based on my real world experience as an IT consultant.  We used to use a SAN setup that also incorporated a server (vm or physical is fine) that ran a software called AppAssure Replay.  Dell recently (in the last year to year and a half) purchased AppAssure and is making them more robust as we speak.  They do, however, provide a really good backup solution already, with a feature built in ( you have to turn it on and configure it which isn't hard) called Live Recovery which basically exports a live working VM of your server(s) to a designated storage spot (preferably a local or offsite SAN) every 5 minutes.  Now, keep in mind that the software is already backing up, deduplicating, and can replicate all backups to another Core (that's what AppAssure calls the actual backup servers with the console installed on them) all on top of this Live Recovery option.  You don't need a SAN as long as you have enough room to wherever you are replicating the Live Recovery and the backups.  They also have a feature called Seeding that allows you to take the initial backup offsite with an external hard drive and that way all 100GBs don't have to go across the Internet, just the changes do.

Let me know if you're interested and I'll set you up with a more detailed plan of what we used to do.

Experts Exchange is like having an extremely knowledgeable team sitting and waiting for your call. Couldn't do my job half as well as I do without it!
James Murphy

Firstly, forgive the ridiculousness of this question.
No problem, those questions are way more fun ;-)

I agree this is a slightly odd request, but what the hell. if this is what they want there's no harm in considering it.

OK, so both Hyper- & VMware have replication facilities built in and can replicate VMs very well if setup accordingly. it's well worth considering but there are a few issues with this option:
Can be expensive to setup
Can slow VMs down if the storage isn't fast enough (uses delta files for VMs which can reduce performance a little)
Is rubbish if you cannot exclude the VM disk that contains the pagefile

Alternatively, you could use a backup facility that can be restored directly to a VM (eg backup exec & many others) which is easier to setup, but this tends to be a little harsh on the bandwidth (which may be an issue for you)

You could consider replicating the applications instead of the VMs (eg exchange DAG, SQL log shipping, DFS file replication etc etc) but this can be costly for purchasing licenses etc, and may not be possible with some applications.

Let me know if any of these are worth discussing in more details and I'm happy to expand on it a bit if you'd like.

Sorry for the absence guys.

Appreciate the input. Do any of these proposals incorporate an automatic failover?

Just to clarify, we will have two completely seperate sites with 1 server on each site. The servers only have internal direct attached storage. There is no budget for properly redundant SAN, and we want to avoid the ipod "Inverted Pyramid of Doom".

All we need is for a backup of a live server to be replicated over a WAN link, and upon failure of the onsite server the DR server has a recent operating copy of the failed system.

It doesnt necessarily have to be "live" DR. But clost as possible with a solution that is bandwidth friendly
Gerald Connolly

Log in or sign up to see answer
Become an EE member today7-DAY FREE TRIAL
Members can start a 7-Day Free trial then enjoy unlimited access to the platform
Sign up - Free for 7 days
Learn why we charge membership fees
We get it - no one likes a content blocker. Take one extra minute and find out why we block content.
See how we're fighting big data
Not exactly the question you had in mind?
Sign up for an EE membership and get your own personalized solution. With an EE membership, you can ask unlimited troubleshooting, research, or opinion questions.
ask a question
Log in to continue reading
Log In
Sign up - Free for 7 days
Get an unlimited membership to EE for less than $4 a week.
Unlimited question asking, solutions, articles and more.

Thanks guys.

Yes, as suspected, there is no good or easy way to do this.  The best answer is no when the closest answer to yes is maybe. The cost of maintaining two WAN links with the appropriate amount of bandwidth to accomodate the requested system is simply unjustifiable. A few hours downtime while a Backup is restored is the best option.

Instead, we are looking into an online backup solution that enables your backup to be immediately loaded as a virtual machine so you have access to the resource even whilst the restoration takes place.

Cheers, Paul
Get an unlimited membership to EE for less than $4 a week.
Unlimited question asking, solutions, articles and more.


StorageCraft Cloud Premium will do what you ask. It snapshots your VMs, creates continuing differentials and allows you to upload all of your backup images to StorageCraft's cloud service. Then, if disaster strikes, you can boot your uploaded images (it consolidates your differentials with your last full snapshot, then boots off of the consolidated image) in the cloud, on their servers.

Now all your servers are running as VMs in their cloud, based off of your backups. Does that make sense?

There are several ways of doing instant failover, but a very cheap one that immediately springs to mind is dynDNS. I.e., have one of your servers update a dynDNS service every minute, so that your hostname(s) point to that IP.

Then have an external service (or rent a virtual host somewhere) ping your DNS hostnames every minute. If it stops responding, send an email to yourself (not to your corporate email account, of course). Or send an email to an email-to-SMS service, so you get a text message on your phone.

After 5 minutes of no response from ping, and no intervention by you, have it kick StorageCraft's DR service into action, so it boots your backup images. All of this can be relatively easily scripted on either Windows or Linux.

Finally, have the script disable dynDNS updates from your original IP once the DR site is online and has registered its IP. You don't want your DNS flip-flopping back and forth every 30 seconds.

How's that for a (relatively) cheap DR solution with failover?

Regards, G
(No connection to StorageCraft, but I do work for a company which has clients who use it, and are happy with it)