SQL Server Licensing: High Availability and Disaster Recovery benefits

[Amit Banerjee]

One of the most popular software assurance benefits for SQL Server customers is the ability to deploy SQL Server High Availability and Disaster Recovery architectures with the best TCO. Starting November 1st, 2019, every Software Assurance customer of SQL Server will be able to use three enhanced benefits for any SQL Server release that is still supported by Microsoft:

• Failover servers for high availability – Allows customers to install and run passive SQL Server instances in a separate operating system environment (OSE) or server for high availability on-premises in anticipation of a failover event. Today, Software Assurance customers have one free passive instance for either high availability or DR
• Failover servers for disaster recovery ^NEW – Allows customers to install and run passive SQL Server instances in a separate OSE or server on-premises for disaster recovery in anticipation of a failover event
• Failover servers for disaster recovery in Azure ^NEW – Allows customers to install and run passive SQL Server instances in a separate OSE or server for disaster recovery in Azure in anticipation of a failover event

In the video below, you will see how the above benefits can help you deploy recommended HADR architectures using SQL Server at the best TCO. 

References:

Blog post about the HADR benefits announcement

SQL Server 2019 License Guide: https://aka.ms/sql2019licenseguide 

[TrevorKohlman]

Hi Amit Banerjee  thanks for the overview.  Do you know of any Guides and Walk though for working for SQL 2019 and the HA/DR and Azure DR setup.  I am trying better to understand this and the setup for this in what is all involved.  As we have many clients that would be interested in this.  My setup would be VM on Prem with Primary, Secondary and Secondary DR.  As well as VM on Prem with Primary , Secondary and Azure Sencondary DR. I want to understand the Testing that you talked about and what we can and can not do under the Software assurance.

[SaraGonc]

Hi Amit Banerjee, thanks for the excelent overview.

I have the same interest of @TrevorKohlman too.

Thanks for your help.

[sybadm]

hey Amit, Nice examples. When you say Software Assurance allows you to run DR Test. Could you please clarify this topic in more details. Does Software Assurance allows to switch Production and DR roles between two sites or this just allows you to test DR and then failback to Production? If later is the case then Software Assurance has no real meaning or any benefits to the customers.

[petervee]

Hi Amit, 

Thanks for this, and other blog posts, @Amit Banerjee 

What would be the requirements for SQL Server (EE/SA) when using AlwaysOn Availability Groups with WSFC on Azure VM's, with no readable-secondaries.  

For example, Azure VM with SQL Server (EE/SA) with 16 cores for primary, and secondary with 16 cores (no readable secondary, active/passive).  Would the benefit cover this scenario (no on-prem)?  Or, would all 32 cores/vCpus need to be licensed?  

Thanks!

[banerjeeamit]

@TrevorKohlman & @SaraGonc The testing refers to any DR drills that you would perform periodically to determine that your DR plan is valid and it will work in case of a true disaster. Please talk to your Account Executive or Licensing Specialist to determine if your specific DR test scenario qualifies for the benefit if you want to validate specifically for your environments. All the SQL Server HADR common options are listed here for Azure Virtual Machine.

From our Product Terms for SQL Server Software Assurance customers:

Customer may also run Primary Workloads and its disaster recovery Fail-over OSEs simultaneously for brief periods of disaster recovery testing every 90 days, and around the time of a disaster, for a brief period, to assist in the transfer between them.

@sybadm For the first part of your Q, please refer the above comment. When you switch between production and DR environments, you are leveraging your License Mobility Software Assurance benefit and reassigning licenses from production to DR. You can do this reassignment any number of times with Software Assurance. 

@petervee For your example (Azure VM with SQL Server (EE/SA) with 16 cores for primary, and secondary with 16 cores (no readable secondary, active/passive).  Would the benefit cover this scenario (no on-prem)?  Or, would all 32 cores/vCpus need to be licensed?):

a. If you are running this setup on Azure VM or on-premises, you would not need to license the secondary replica as long as it has 16 cores or less and the replica satisfies the condition of a passive replica. 

b. The benefit that you would leverage would be the fail-over server Software Assurance benefit. If you add a third replica with 16 cores which is passive but is set for async data movement, then that would not need to be licensed under the new failover server DR benefit. 

c. If you are running on-premises, and set up a fourth async passive replica on Azure VM for 16 cores, then you wouldn't need to license the SQL licenses on the VM due to the Failover servers for disaster recovery in Azure benefit. 

HTH

[Martin76]

@Amit Banerjee 

as per new licenses SA benefits, I know now one licenses is enough for  this scenarios.  however I want to confirm below as earlier  almost all customers believed  SQL Log shipping for DR in this scenarios licenses free.   

Primary Site  HA( SQL Failover Cluster Instance - Active/Passive - Not Always on ), If it is always on, agree they need licenses for DR as DR copy is secondary copy 

DR Site ( Standalone Passive Instance - SQL Log shipping ) 

Did they need have 2 Licenses( old license rules )

old version of SQL , most of customers use above scenario to save  licenses for DR as it is not active unless for DR.  

If Customer use Primary site ( always on for HA - Primary Copy in Second Server) - license was not required for primary copy and DR  site always on  secondary Copy ( or Log shipping ) need licenses ( old licensing rule ). this is clear. 

[IvanKobe]

Hi @Amit Banerjee

Glad I found these videos!  Really appreciate you creating and posting them, as they go a long way in helping understand the SQL licensing rules around HADR!  I do want to confirm however, as going through various points of reference around Standard vs Enterprise editions, has left me wondering and am hoping you can clarify!

I currently have a Primary instance of Standard SQL 2019 w/SA, running in a VM with 4 cores assigned to it.  I have licensed all 4-cores (with dual 2-core packs).  Within this instance I have 8 individual databases. 

My first question is... with the new HADR benefits that you've described in the videos above... does it pertain to this instance with all 8 individual databases, or is there a limit to how many databases are supported under a Standard Edition (w/SA) Instance?

Second question is... understanding that we can have 3 PASSIVE secondary's (for HA, DR and DRinAz)... do the secondary's need Windows Server licenses, or does Mobility licensing come into play, since these are not active OSEs?

Thanks!

[Sriramang]

Dear Amit,

I have gone through your Video and in our Scenario which we would like to explain, kindly provide us the answers:

a. We have 2 Servers ,  DC1 - Server 1 - DC2  - Server 2

     We would be installing SQL server Standard in each  of them

b. Each Server has 2 CPU ( Core) 

Question:

c.   The Servers will have real time updates 

d.  The DC2 server has to take over the operations in case Failure at DC1.

c. Could you advise us on the licensing ?

d. Any additional software required for Data Replication between DC1 & DC 2

e. Read Write application should be simultaneous at the same time always.

f. When DC1 comes back alive, operations shall shift back to DC 1

Could you help us with answers

regards

[Joe Rainero]

Is this for SQL Standard or SQL Enterprise licensing?

[IvanPKobe]

@Joe Rainero  The information applies both for SQL Standard as well as Enterprise Licensing.  That being said, there are some limitations with Standard that must be accounted for, in where for Enterprise are not encountered.

[IvanPKobe]

@Sriramang 

With regards to your scenario and questions, I have provided answers (based upon the videos above).

a. We have 2 Servers ,  DC1 - Server 1 - DC2  - Server 2

     We would be installing SQL server Standard in each  of them

IKe Answer: Sure this is a good starting point.

b. Each Server has 2 CPU ( Core) 

IKe Answer:  First off, let's clarify what you mean by Each Server has 2 CPU (Core).  I am going to assume that what you mean is, you have 2 x VM's... one VM on DC1 and a second VM on DC2.  Each VM you are assigning 2 vCPU's.  Hopefully this interpretation is correct.  Please clarify if not.  

So with that in mind... my recommendation is to create a VM with a minimum of 4 vCPU's.  The reason is you will have to license a minimum of 4-cores irrespective if you only start with 2 vCPU's.  (This is a SQL licensing requirement)

c.   The Servers will have real time updates 

IKe Answer:  Okee dokee, as I'm not really sure as to what you mean by this.

d.  The DC2 server has to take over the operations in case Failure at DC1.

IKe Answer:  So from the perspective of HADR that is being exemplified in the video, you will have DC1 being your Primary SQL Instance.  DC2 will be your Secondary SQL Instance.  They will be clustered together via Windows Server Failover Cluster (and as such you will need a Quorum Witness).  From there you will create a Basic Availability Group for EACH database you have on your SQL server that you want to have HADR for.  (If you were using Enterprise licensing, you would only have to create a single Availability Group).  So on the failure of DC1, DC2 would AUTOMATICALLY take over the role of Primary SQL server.

c. Could you advise us on the licensing ?

IKe Answer: Under the HADR schema, you would need 2 x SQL Standard (2-core) licenses.

d. Any additional software required for Data Replication between DC1 & DC 2

IKe Answer: As mentioned above, you will be creating a Windows Server Failover Cluster.  I would recommend reading up on the requirements of implementing.

e. Read Write application should be simultaneous at the same time always.

IKe Answer:  This would be negative.  What you are implementing here via the above videos is an HA or DR scenario.  In both cases, you will have a single Primary Instance and then multiple Secondary Instances.  The video does not go about talking above a multiple Primary SQL instance solution.  Therefore, Read/Write applications will only always go to the Primary SQL only.

f. When DC1 comes back alive, operations shall shift back to DC 1

IKe Answer:  Totally your choice.  Manual failback can be pushed out so that DC1 is always your Primary instance, and thereby causes less confusion with any troubleshooting.

[Adal3088]

Hi, can you guide me regarding the passive replicas of HA and DR, in my primary server I have several jobs, these must be replicated in my HA and DR to be prepared for a failover, when doing this and my jobs are executed in the HA and DR while they are still passive even if they are executed with errors, does it imply acquiring licenses for these passives?

Greetings.

[AnabilSaha]

Hello Mr Banerjee. It's a really very helpful video. I have a real time scenario can you please suggest me what is your view on that ?

In my current production server an architecture was decided that is going to have 2 vms for db server where 1 will be primary and 2nd will be secondary. But suddenly I am being forced to change the architecture like this. Primary will be on physical rac server and secondary will be on VM server. So my questions are following - 

1. My company have purchased 2 2 core based enterprise edition licence. How is it going to work for this scenario.

2. Any complications will arise during setup HA 

3. Any other suggestions from you.

Thanks and Regards 

Anabil Saha