If you follow my guide, go to “Should run on hosts in the group” or “Should not work on hosts in the group” unless you can`t avoid it. If you don`t have special license agreements that prohibit you from migrating a particular application to different hosts, I don`t see a good reason to use one of the other two rules available. As you can see, the two types of rules we can create are perfect and adapt to most scenarios. As mentioned, these types of policies are typically required for storage appliances or backup appliances that work closely with the host. These virtual machines are not meant to be moved because they typically use the host`s physical disks. Consider the differences between “shall” and “should.” You can control the placement of virtual machines on hosts within a cluster by using affinity rules. Many new administrators who are new to VMware technology do not know how to properly use VM/Host rules for affinity and anti-affinity. We hope that this article has helped to dispel some doubts. I`d like to know why you use “Must work on group hosts” or “Must not work on group hosts” instead of the other two rules. I`ve sometimes heard VMware administrators explain that VM-Host affinity rules should always be preferred, not mandatory. The reason is simple.
For example, suppose one of the hosts in a host group crashes or disconnects for maintenance, and the remaining hosts in the host group do not have sufficient resources to fully absorb the virtual machines running on the host that are currently running offline. If the VM host affinity rule states that these VMs run only on hosts in the specified host group, the rule prevents at least some of the VMs from running, even if hardware resources are available in other host groups. The “Must run on hosts in the group” rules can be useful if your virtual machines require specialized hardware, such as GPUs in the hosts. Excellent post Frank, it`s interesting to see how binding rules can restrict and affect HA/DRS, and we should make sure this is factored into our plans. Keep up the good work!! How does HA handle preference rules? VMware High Availability follows binding rules and binding rules when placing virtual machines after host failover. Only virtual machines specified in the DRS host group can be placed on ESX hosts. DRS does not inform HA of the existence of preference rules, so HA is not aware of these rules. High availability cannot prevent the virtual machine from being placed on an ESX host that is not part of the DRS host group, in violation of the affinity rule. DRS will correct this violation on the next call. How does the SRD manage preferential rules? During a DRS call, DRS executes the algorithm with preferred rules as binding rules and evaluates the result. If the result contains violations of cluster restrictions; If you overbook a host or overload a host, resulting in 100% CPU or memory usage, preferential rules are ignored and the algorithm is run again. Essentially, a VM-to-host affinity rule limits the number of hosts on which virtual machines can be powered on or migrated.
Setting VM-to-host affinity rules can limit load balancing and evacuation for maintenance mode. Load balancing limitations Some risks are introduced when mandatory VM-host affinity rules are used. Because of the restrictive behavior that allows virtual machines to start only on the ESX host associated with the DRS host group, this affects HA`s ability to select a compatible ESX host to place the virtual machine. In addition, using mandatory VM-to-host affinity rules reduces the virtual machine placement options used by DRS when defragmenting the cluster. Using percentage-based high availability can lead to fragmentation of the access control resource. During a failover, HA requests DRS defragmentation. DRS attempts to migrate virtual machines to reclaim enough unfragmented resources for all virtual machines. Because DRS is allowed to use “multi-hop” migrations, DRS calculations typically create a “chain” of migrations when defragmenting a host.
For example, VM-A migrates to host 2 and VM-B migrates from host 2 to host 3. Mandatory rules restrict the rules of the game and allow virtual machines to move only within the associated DRS host group, reducing the overall options for transporting virtual machines through the cluster, regardless of the mapping to VM-host affinity rules. DRS maintenance mode does not violate CPU and memory allocation to comply with mandatory VM-host affinity rules, and it does not violate mandatory rules for reservations to be honored. When placement, both conditions must be met, and therefore DRS only places a virtual machine if its reservation and mandatory rule can be met. This behavior affects DRS`s ability to select an appropriate compatible host to place virtual machines during automatic evacuation in maintenance mode. Conclusion Now it`s up to you to decide which rule is appropriate for distributing workloads among the ESX hosts in the cluster. Knowing the implications and limitations introduced by mandatory regulations can help you make an informed decision. The use of binding affinity rules can sometimes also help control licensing costs. Some application providers use a host-level licensing model that requires a host server that can host the licensed application. Therefore, creating a small host group and configuring an affinity rule so that the application server can run only on hosts in that group would allow the organization to avoid the cost of licensing additional hosts to run the application. Providing these two options creates a new problem for the administratorarchitect, when will the need to use the mandatory rule arise, and when is it desirable to use the preferred rules? I think it all depends on the risks and limitations that each rule brings.
Let`s look at the difference between the rules, the behavior of each rule, and the impact on cluster services and maintenance mode. What is the difference between a mandatory rule and a rule of preference? Frank, excellent contribution. I just wanted to restrict a VM to an ESX host (these are SLR firewall VMs). In this case, I set up “must” rules. What happens with the ESX host crash? VMware Distributed Resource Scheduler (DRS) balances and optimizes the performance and DRS score of the virtual machine (the DRS score is a measure of the resources that can be used by the virtual machine). The higher the DRS value of a virtual machine, the better the availability of resources for that virtual machine. VMware DRS moves a virtual machine from one host to another to enhance this virtual machine DRS value. However, it adheres to VM-to-VM affinity and anti-affinity rules. If you add or modify an affinity rule and the current state of the cluster violates the rule, the system continues to function and attempts to correct the violation. For manual and semi-automated DRS clusters, migration recommendations based on policy execution and load balancing are submitted for approval. You are not required to comply with the rules, but the relevant recommendations will remain in place until the rules are followed.
To verify that the affinity rules enabled are not respected and cannot be corrected by DRS, select the DRS tab of the cluster and click Errors.