FCI Alignment

Vishu 1,516 Reputation points
2021-06-22T13:39:50.117+00:00

Experts,

A Generalized question , if I have a 2 node cluster with FCI instances active on both nodes , one active instance on each node.

8 cores 156 GB on each node. 64 GB aligned to each instance on their respective nodes.

From memory perspective I understand total I will be using 128(64+64) GB if both Instances failover to one node. However just on a general basis if the CPU utilization is 40% on each node , when both the Instances failover to one node , am I liable to get performance issues.

There are no performance issues when instances on their indusial nodes. Just wanted to understand if there could performance issues due to failover on one node on a high level , though I understand there could be various other factors also.

SQL Server
SQL Server
A family of Microsoft relational database management and analysis systems for e-commerce, line-of-business, and data warehousing solutions.
12,605 questions
0 comments
Accepted answer
  1. Jeffrey Williams 1,886 Reputation points
    2021-06-22T19:08:48.8+00:00

    There could be performance issues - but if both instances never go above 40% utilization then I wouldn't think performance will suffer. With that said - I would be concerned because there are going to be times on each instance where they spike to 60%+ which can and will affect the other instance.

    0 comments

1 additional answer

Sort by: Most helpful
Most helpful Newest Oldest
  1. Seeya Xi-MSFT 16,436 Reputation points
    2021-06-23T06:44:53.017+00:00

    Hi @Vishu ,

    Based on your description, the topology you are using is an active/active cluster. And I agree with Jeffrey.

    Advantages: The hardware resources of both nodes can be fully utilized (two or more SQLSERVER cluster instances need to be installed), and cost savings.
    Disadvantages: Once a node fails over, there will be two instances of SQLSERVER running on another node at the same time. At this time, the two instances may contend for resources such as CPU, memory, I/O, etc. on this node, causing the performance of both instances to be affected. Sometimes users on both sides may not accept it. Therefore, the problem on the abnormal node must be resolved as soon as possible, and the failover instance must be switched back as soon as possible.

    In addition to "active/inactive" clusters and active/active clusters, I will introduce another topology for you, that is, a cluster of N active + 1 inactive nodes. Take a 3-node cluster as an example. Install two SQLSERVER cluster instances on it. The Possible Owner of each instance contains two nodes in the cluster, but only one node is shared by both instances. Under normal conditions, the two SQLSERVERs are running on the non-shared node and are not related to each other. Once a node fails over, it will switch to that shared inactive node.
    This structure is a scheme between "active/inactive" and "active/active". Compared with "active/inactive", he wastes less node resources (1/N+1). In addition, if two or more nodes fail over at the same time, the probability of switching to a shared node at the same time is relatively low, so the performance problem of the "active/active" structure is solved to a certain extent.
    The picture is as follows:
    108453-n1.png

    Best regards,
    Seeya

    If the response is helpful, please click "Accept Answer" and upvote it, as this could help other community members looking for similar queries.
    Note: Please follow the steps in our documentation to enable e-mail notifications if you want to receive the related email notification for this thread.

    0 comments