Hyper-V replication licensing
-
Or just buy a second license and do whatever the hell you want.
Or, like probably like 95% of people do, say "That's a bunch of crap" and just do what you want.
-
I never expected the thread to blow up while I was away. I meant two servers are needed for the remote office. A domain controller and a file server. Corporate IT insists that the DC and FS be physical. Corporate says only physical servers are allowed because virtual servers are too risky. Since they obviously aren't following best practices, I was trying to make a business case out of virtualization and wondered if we could save $600 on the license of Windows. I was proposing taking their hardware and only buying one Windows license to come out $600 ahead and have a redundant system instead of having production go down if either one of their servers fails.
-
I agree that in the big picture of their infrastructure $600 isn't much, but this has been an interesting group to work with. I could share a few examples of some of the stuff that they said over the last couple of days, but I don't need to berate anyone, I need to have a talk with the business manager and show them how their IT is making bad business decisions.
-
@scottalanmiller said in Hyper-V replication licensing:
@Dashrender said in Hyper-V replication licensing:
@scottalanmiller said in Hyper-V replication licensing:
@Dashrender said in Hyper-V replication licensing:
@scottalanmiller said in Hyper-V replication licensing:
@Dashrender said in Hyper-V replication licensing:
But I have to ask, if you are dropping a few grand for the hardware, why would you not spend the extra on the second License and have HA setup?
Because you can have HA without spending that money AND in most SMBs that's still a lot of extra money.
So what... You setup autofailover, then when you see a failover you then disable that until the 90 days is over?
Yup, that's one way. If you are doing Hyper-V Replication, though, there isn't any means of failing back as the replication is one way, so nothing more to do.
Really? So once it's failed, and you repair the other box, wait 90 days, then you have to configure replication back the other way?
Yup. That's how one way replication works
Yes, it may be technically how it works.
But no, that is not how it works as a system. The system handles it all for you.Note that the primary and replica servers are now reversed.
-
@JaredBusch Thanks for taking the time to go in to detail of how that works. Right now all my larger systems are VMware so I haven't done anything large with Hyper-V. I can see the writing on the wall though and I think my next cluster will be Hyper-V.
-
For a failed system failover, it works like this.
I manually shut down the live one and know that all my data is replicated so I am not losing anything for this.
In this case you have to manually reverse the replication once the failed host is back up.
-
@Mike-Davis said in Hyper-V replication licensing:
I never expected the thread to blow up while I was away. I meant two servers are needed for the remote office. A domain controller and a file server. Corporate IT insists that the DC and FS be physical. Corporate says only physical servers are allowed because virtual servers are too risky. Since they obviously aren't following best practices, I was trying to make a business case out of virtualization and wondered if we could save $600 on the license of Windows. I was proposing taking their hardware and only buying one Windows license to come out $600 ahead and have a redundant system instead of having production go down if either one of their servers fails.
Not only can you save $700, but you can provide failover, too, which is completely missing from their "low risk" scenario. Plus you can snapshot before patching, further protecting them from themselves.
But if they are mandating physical, does risk or cost savings really come into play?
-
@Mike-Davis said in Hyper-V replication licensing:
I agree that in the big picture of their infrastructure $600 isn't much, but this has been an interesting group to work with. I could share a few examples of some of the stuff that they said over the last couple of days, but I don't need to berate anyone, I need to have a talk with the business manager and show them how their IT is making bad business decisions.
Where "bad" = "professional negligence."
-
But for a mere $600 more you can provide true High availability with fail over capabilities between two servers which they are already going to buy..
-
@DustinB3403 said in Hyper-V replication licensing:
But for a mere $600 more you can provide true High availability with fail over capabilities between two servers which they are already going to buy..
Define "true HA" here? If you define it by the total environment, it's not really a factor. If you define it by the ability to failover, again, we covered, not a real factor.
How does it impact downtime given that you can have another license almost as quickly as you can restore a system?
-
Wouldn't there also be a hardware savings?
-
True HA at the OS level would be the ability to migrate the VMs on whim. This would mean obviously having licensing in place on every possible host for the potential total capacity of the server.
This could very easily put you into a Datacenter license, but in many cases would be a mere 2 server standard licebses.
-
@BRRABill said in Hyper-V replication licensing:
Wouldn't there also be a hardware savings?
No because without a second host they have no DR fail over plan at all. Besides restoring from backup.
-
@BRRABill said in Hyper-V replication licensing:
Wouldn't there also be a hardware savings?
From which aspect? Do you mean virtualizing? In theory, yes, they could, instead of HA, go down to one server and STILL improve reliability while greatly reducing cost.
-
@DustinB3403 said in Hyper-V replication licensing:
@BRRABill said in Hyper-V replication licensing:
Wouldn't there also be a hardware savings?
No because without a second host they have no DR fail over plan at all. Besides restoring from backup.
Not technically any need for the second host, they could just do without.
-
The idea as I understand it what @Mike-Davis haso said was to purchase 2 licenses for bare metal installs from the get-go
-
@DustinB3403 said in Hyper-V replication licensing:
True HA at the OS level would be the ability to migrate the VMs on whim.
That's not what HA refers to. HA refers to uptime. Mainframes have no ability to migrate anything to anywhere, but are extreme HA. Ability to migrate and HA are unrelated concepts. There is an importance to portability, for sure, but it is not directly associated with HA.
-
@scottalanmiller said in Hyper-V replication licensing:
@DustinB3403 said in Hyper-V replication licensing:
True HA at the OS level would be the ability to migrate the VMs on whim.
That's not what HA refers to. HA refers to uptime. Mainframes have no ability to migrate anything to anywhere, but are extreme HA. Ability to migrate and HA are unrelated concepts. There is an importance to portability, for sure, but it is not directly associated with HA.
SMB space scott. HA for an smb ends at 99.9% it literally doesn't go further.
-
@DustinB3403 said in Hyper-V replication licensing:
@scottalanmiller said in Hyper-V replication licensing:
@DustinB3403 said in Hyper-V replication licensing:
True HA at the OS level would be the ability to migrate the VMs on whim.
That's not what HA refers to. HA refers to uptime. Mainframes have no ability to migrate anything to anywhere, but are extreme HA. Ability to migrate and HA are unrelated concepts. There is an importance to portability, for sure, but it is not directly associated with HA.
SMB space scott. HA for an smb ends at 99.9% it literally doesn't go further.
Maybe even 99.0%
-
@DustinB3403 said in Hyper-V replication licensing:
@scottalanmiller said in Hyper-V replication licensing:
@DustinB3403 said in Hyper-V replication licensing:
True HA at the OS level would be the ability to migrate the VMs on whim.
That's not what HA refers to. HA refers to uptime. Mainframes have no ability to migrate anything to anywhere, but are extreme HA. Ability to migrate and HA are unrelated concepts. There is an importance to portability, for sure, but it is not directly associated with HA.
SMB space scott. HA for an smb ends at 99.9% it literally doesn't go further.
Then you get HA without any of this. Normal servers are 99.99 - 99.999% out of the gate. So by that logic, you always have HA.
But that's never what HA means. HA is an order of magnitude or better than SA. Having "HA is lower that SA" makes no sense, no matter how small the business is. The idea that SMB has a different concept of HA doesn't make any sense because the same baseline always exists, and that baseline is 99.99% or higher by default.