March 26, 2012 at 7:12 am
Hi;
I have a active passive cluster (2 node)
Now I want to create a seprate active/active cluster (2 node)
Appreciate if anyone help me to resolve my following query
1 - how many SQL server licences will be reuqired?
2 - how much memory will be available?
3 - how much CPU will be available for me?
4 - in case if 1 node goes down what will be the affect on Mem & CPU?
Detail
my all 3 nodes will have following h/w config
64 GB Ram on each node
2 physical proc on each
Total 3 nodes
March 26, 2012 at 7:30 am
Firstly lose the 'active passive and active/active terms'
What you have is, a 2-node single instance cluster, a single SQL instance installed on a 2-node cluster
What you want to create is a 2-node, multi-instance cluster (2 instances on a 2-node cluster)
Now, where does the 3 nodes come from?
The questions about memory and CPU don't make sense.
Gail Shaw
Microsoft Certified Master: SQL Server, MVP, M.Sc (Comp Sci)
SQL In The Wild: Discussions on DB performance with occasional diversions into recoverability
March 26, 2012 at 7:30 am
SQL_Helper (3/26/2012)
Hi;I have a active passive cluster (2 node)
Now I want to create a seprate active/active cluster (2 node)
Appreciate if anyone help me to resolve my following query
1 - how many SQL server licences will be reuqired?
2 - how much memory will be available?
3 - how much CPU will be available for me?
4 - in case if 1 node goes down what will be the affect on Mem & CPU?
Detail
my all 3 nodes will have following h/w config
64 GB Ram on each node
2 physical proc on each
Total 3 nodes
1 - you will need 4 CPU licences if a 2 node cluster with 2 CPU's in each node running Active/Active
2 - if you have 64GB in a node then you have 64gb available
3 - if you have 2 cpu's then you have 2 cpu's available
4 - if you max each instance out at say 60GB ram and it runs at 90% CPU then if it fails over you will have an issue, if you want it active active then you need to split the resources, assign 30GB ram to each instance, that way if the two isntance are running on 1 node you have used 60GB with 4GB for the OS, you can do the same with CPU affinities but if total CPU of both instances running together is lower than 80% then you should be ok without playing with the affinities.
March 27, 2012 at 12:02 am
Hi Champ;
First of all thanks for reply, I think I was not able to express my query properly. let me explain Mathamatically 😉
Available resources:
================
1) Total Servers available = 3
2) RAM on each server = 64GB
3) Physical CPU on each server = 2
4) Total Core in each CPU = 4
I am going to create 3 node cluster (2 active nodes and 1 passive node)
Q1 - how much total memory will be availabe?
Q2 - how much CPU will be available?
Q3 - if 1 active node will goes down, what will be the case?
Q4 - if 2 nodes down, what will be the case?
I hope above information will be helpfull to explain my objective, appreciate if anyone response to implement this infrastructure, thanks.
March 27, 2012 at 12:13 am
Available resources:
================
1) Total Servers available = 3
2) RAM on each server = 64GB
3) Physical CPU on each server = 2
4) Total Core in each CPU = 4
I am going to create 3 node cluster (2 active nodes and 1 passive node)
Q1 - how much total memory will be availabe?
Q2 - how much CPU will be available?
Q3 - if 1 active node will goes down, what will be the case?
Q4 - if 2 nodes down, what will be the case?
The memory of each server will be available. You cant fail over physical cpu and memory resources (if that is what you were asking?)
Q3 - It depends on how you set up your cluster; which node servers as the passive node for this active node?
Once fail over occurs, all resource groups (SQL Services) will fail over to your passive node.
Q4 - See answer to Q3.
One thing to note: When you use 1 Passive node for 2 active nodes, and both fail over to 1 passive node, your performance might degrade as one server now hosts 2 servers resources.
How much, we cant say.
-----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
This thing is addressing problems that dont exist. Its solution-ism at its worst. We are dumbing down machines that are inherently superior. - Gilfoyle
March 27, 2012 at 12:29 am
Hi
Greate, thanks for the response.
I am expanding this discussion 1 step further.
Now we have 3 nodes cluster (2 active and 1 passive), Please help to answer following queries.
Requirement:
==========
a) I want to install 3 instances on my cluster
b) I will assign 50 GB to instance 1
c) I will assign 50 GB to instance 2
d) I will assign 10 GB to instance 1
Q1 - what will be the best plan to install 3 instances on 3 node cluster (2 active and 1 passive)?
Q2 - in case of 1 active node failure, will it transfer instance automatically to passive node or need to transfer manually?
Q3 - in case of 2 active node failure at the same time, will it transfer instaces automatice or need to transfer manually or will it not transfer 2 nodes load to one node?
March 27, 2012 at 12:38 am
Requirement:
==========
a) I want to install 3 instances on my cluster
b) I will assign 50 GB to instance 1
c) I will assign 50 GB to instance 2
d) I will assign 10 GB to instance 1
Q1 - what will be the best plan to install 3 instances on 3 node cluster (2 active and 1 passive)?
Q2 - in case of 1 active node failure, will it transfer instance automatically to passive node or need to transfer manually?
Q3 - in case of 2 active node failure at the same time, will it transfer instaces automatice or need to transfer manually or will it not transfer 2 nodes load to one node?
A1 - Ok, first things first. What is the purpose of your cluster?
A2 - Yes, automatic fail over will occur, Except if you set it to manual, but by default auto fail over will occur.
A3 - If both Active Node Instances fail at the same time, and they were both configured to fail over to the same Passive node, then yes, all instances will run on the passive node, until they are moved back to their original Active Node hosts.
Regarding your Capacity Assignments:
Read this Cluster documentation
http://msdn.microsoft.com/en-us/library/ms179410(v=sql.105).aspx
http://technet.microsoft.com/en-us/library/ms143506(v=sql.100).aspx
-----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
This thing is addressing problems that dont exist. Its solution-ism at its worst. We are dumbing down machines that are inherently superior. - Gilfoyle
March 27, 2012 at 2:14 am
Hi;
Thanks again, now move another step forward
As Now we have 3 nodes (2 active and 1 passive) cluster for our production, Now its time to create another cluster for DR SITE.
This time I am creating 2 node active active cluster for my disaster receovery site.
Resources avialable at DR SITE:
========================
1) Total Servers available = 2
2) RAM on each server = 64GB
3) Physical CPU on each server = 2
4) Total Core in each CPU = 4
The most important thing is this we have SAN to SAN replication b/w Production and DR SAN
Now how can I implement disaster recovery solution
As per the above senario what will be the best approach to implement disaster receovery solution..
March 27, 2012 at 2:21 am
This time I am creating 2 node active active cluster for my disaster receovery site.
As Gail said, when you refer to A/A or A/P clusters, be sure you know what they mean.
Active Active in a Microsoft sense means you are running 2 x 1 Active Nodes, referring to each other as fail over passive nodes.
Personally, I would speak to the SAN Admin's about the replication.
-----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
This thing is addressing problems that dont exist. Its solution-ism at its worst. We are dumbing down machines that are inherently superior. - Gilfoyle
March 27, 2012 at 10:24 am
SQL_Helper (3/27/2012)
HiGreate, thanks for the response.
I am expanding this discussion 1 step further.
Now we have 3 nodes cluster (2 active and 1 passive), Please help to answer following queries.
Requirement:
==========
a) I want to install 3 instances on my cluster
b) I will assign 50 GB to instance 1
c) I will assign 50 GB to instance 2
d) I will assign 10 GB to instance 1
Q1 - what will be the best plan to install 3 instances on 3 node cluster (2 active and 1 passive)?
Q2 - in case of 1 active node failure, will it transfer instance automatically to passive node or need to transfer manually?
Q3 - in case of 2 active node failure at the same time, will it transfer instaces automatice or need to transfer manually or will it not transfer 2 nodes load to one node?
It seems to me that you are extremely confused, sit down and carefully workout where you plan to deploy the instances.
For your 3 node cluster, the following would be a typical install scenario;
NodeA
Install sql server instance 1
NodeB
Install sql server instance 2 and sql server instance 3
NodeC
Install sql server instance 1, 2 and 3
On any ordinary day NodeC would be totally idle, should a failure occur on NodeA then instance 1 would fail over to NodeC. Similarly, any failure on NodeB would cause instances 2 and 3 to fail over to NodeC.
As long as the fail over period is less than the 28 days grace provided by the SQL Server licensing you would not need to licence SQL Server on NodeC, the Windows OS however would need to be licenced.
Is this the design you are intending to implement?
On a failover cluster generally the fail over is automatic although you can actually control this to a certain degree, what is the level of fail over you require?
-----------------------------------------------------------------------------------------------------------
"Ya can't make an omelette without breaking just a few eggs" 😉
Viewing 10 posts - 1 through 9 (of 9 total)
You must be logged in to reply to this topic. Login to reply