How to Create and Manage Data Centers and Clusters in RedHat Virtualization
Introduction to Data Centers
The top-level organizational object in Red Hat Virtualization is the data center. A data center contains all the physical and logical resources in a single managed virtual environment. It is a collection of resources, including clusters, hosts, logical networks, and storage domains.
A single Red Hat Virtualization data center is a self-contained virtualization environment. It may consist of:
- Resources that are all in a particular physical data center at a particular location
- A set of systems and storage belonging to a particular business unit of the organization
- Some other arbitrary division or organization which makes sense to the administrator
A data center can be used to isolate resources belonging to an organization or group from other organizations or groups that should not have access to those resources. This feature allows you to restrict access to data and servers to a specific user group.
One characteristic of a Red Hat Virtualization data center is that all hosts and clusters in that data center share the same storage and must be able to access that storage. Therefore, if some hosts can not or should not be configured to access certain storage resources, then those hosts need to be in clusters in a separate Red Hat Virtualization data center.
A data center named Default is created automatically. Additional data centers can be created, although for many sites there may be no reason to create additional data centers beyond the initial default.
Creating a new data center
The installation of Red Hat Virtualization creates a default data center, named Default. Additional data centers can be created using the Administration Portal. The following procedure details how to create a new data center by using the Administration Portal while logged in as the admin user. This procedure creates a data center that does not yet have any resources associated with it, such as storage domains or clusters. Those resources can be associated with the data center later.
1. Click the Data Centers tab to display the options available to manage data centers. By default, a data center named Default is created when you install Red Hat Virtualization. Initially, theDefault data center is empty and has no resources assigned. Resources like storage and hosts can be assigned to the data center after you create it.
2. Click New to create a new data center. In the pop-up window, titled New Data Center, enter the name you want to use for the data center in the Name field.
3. Select the storage type to use in the new data center using the Storage Type menu. There are two options available: Shared and Local. In most cases, you should select Shared, which allows the data center to contain multiple clusters and hosts that can run virtual machines. If you select Local, the data center is restricted to having a single cluster consisting of exactly one host, but the data center’s storage may be provided by that host’s local file system.
4. Select the Red Hat Virtualization Compatibility Version supported by the data center. For a new data center, you should generally select the latest version available.
When Red Hat Virtualization is upgraded to a newer version, existing data centers, clusters, and hosts may still be configured to use an older version of the product. Any existing hosts and clusters assigned to the data center need to be able to support the selected compatibility version. This ensures that all clusters in the data center support a particular set of Red Hat Virtualization features.
5. Red Hat Virtualization supports quotas that you may use to limit the usage of memory, CPU, and storage resources. A data center can be configured to use these quota settings using the Quota Mode menu.
There are three options available: Disabled, Audit, and Enforced:
- Disabled turns off quota-based restrictions
- Audit allows you to set quota limits but does not enforce them
- Enforced restricts resource use based on quota settings
Like most data center settings, this can be adjusted after you create the data center. When done, click OK. A new pop-up window, titled Data Center – Guide Me appears.
6. The Data Center – Guide Me wizard provides an easy way to configure the other resources that must be assigned to the data center in order for it to be useful. This includes hosts that run virtual machines, clusters to organize those hosts, and storage domains for virtual machine disks and installation media. Each button opens up a new wizard to configure each of these resources.
If you are setting up a new data center, those resources might not yet be set up. In that case, you can click the Configure Later button to complete the configuration of the data center later.
7. When first created, your new data center’s status is Uninitialized. This status changes to Up when the resources are assigned to the data center and Red Hat Virtualization confirms that the data center can use them.
NOTE: Information on how to configure resources and assign them to an existing data center are covered in upcoming sections of this course.
Introduction to Clusters
A cluster is a group of hosts that share the same architecture and CPU model and belong to a particular data center. A cluster is also a migration domain for virtual machines. That means that virtual machines can only be live-migrated between hosts that are in the same cluster.
If a cluster mixes hosts that have different CPU models, the entire cluster is downgraded to match a virtual CPU model that only includes features common to all hardware in the cluster. This helps ensure that virtual machines can be live-migrated cleanly between all hosts in the cluster.
However, this means that virtual machine code may not be able to take advantage of newer CPU instructions that would otherwise be available on hosts in the cluster that have a “better” CPU model. This can reduce virtual machine performance. For this reason, all hosts assigned to the same cluster should have the same CPU model.
Which CPU type the cluster uses is a configuration setting. The cluster’s CPU type must be set so that all hosts assigned to the cluster can provide the features of that CPU type. If a host cannot provide features required by the cluster’s CPU type, it is not available to the cluster.
NOTE: Red Hat recommends that you standardize the make, model, hardware, and firmware or BIOS of all hosts which are assigned to the same cluster. This helps ensure that virtual machines in the cluster have predictable performance characteristics no matter which host they run on, assuming the same workload and set of virtual machines.
A data center may have multiple clusters. You can use clusters to segregate different types of hardware into classes or groups. Another use is to segregate different application components from each other. For example, the virtual machines running the front-end of an application might run in one cluster while back-end virtual machines run in a separate cluster in order to naturally segregate their workloads.
NOTE:All clusters in the same data center share the same storage domains. This is why a data center configured to use Local storage is restricted to having one cluster and one host because only that host can see its local storage. This restriction also means that a data center using local storage cannot use features like live migration, and generally makes it less suitable for production uses.
Creating a new cluster
After installation, Red Hat Virtualization creates an empty cluster named Default in the initial Default data center. Additional clusters can be created using the Administration Portal.
The following procedure details how to use the Administration Portal to create a new cluster in an existing data center. It assumes that you are logged in as the admin user or some other user with equivalent privileges. After the cluster is created, hosts can be added to it and removed from it as needed.
1. Click the Hosts tab to display the hosts available. Select a host in the list, and in the General tab, click Hardware. The hardware specifications for the host are listed. The CPU Type field is a required value to create the cluster because all the hosts in a cluster need to have the same CPU type.
2. Click the Clusters tab to display the options available to manage clusters. Click New to create a new cluster. A pop-up window, titled New Cluster, opens.
3. In the New Cluster window, ensure that the General section is highlighted by clicking on it. Use the Data Center menu to select which data center the new cluster should belong to.
4. Configure the name for the cluster using the Name field. Optionally enter a description or a comment for the cluster using the Description and Comment fields.
5. From the Management Network menu, select the network that the cluster will use for management traffic. If no other networks are configured, it will be the default network for all traffic.
6. Select the CPU architecture used by the hosts in the cluster using the CPU Architecture menu. This CPU architecture is the same for all hosts in the cluster.
7. Select the CPU type used by the hosts in the cluster using the CPU Type menu. This CPU type is the same for all hosts in the cluster.
Intel 64 and AMD64 (x86_64 architecture) CPU types are listed in order from oldest to newest. If a CPU type is selected that requires features not supported by a host in the cluster, that host will not function. Therefore, select a CPU type that matches the features provided by all hosts in the cluster. Another way to describe this is that the CPU type should match the “oldest” CPU or the one with the fewest features in the cluster.
8. Select the Compatibility Version of Red Hat Virtualization that the cluster should use. For a new cluster, you should normally choose the most recent version.
This controls the protocols and features that Red Hat Virtualization uses when managing this cluster. You cannot select a version lower than the Compatibility Version of the cluster’s data center. This is used when upgrading your environment to the latest version of Red Hat Virtualization without disrupting an operating environment.
9. Select the network Switch type used by the cluster. There are two switch types available: Linux Bridge and OVS (experimental). In most cases, you should select Linux Bridge.
10. The check box Enable Virt Service allows hosts in this cluster to run virtual machines, and you normally want to select it.
The check box Enable Gluster Service is used for special clusters that provide GlusterFS service, and in the basic case, it should not be selected.
The Enable to set VM maintenance reason and Enable to set Host maintenance reason checkboxes enable an optional “reason” field when a virtual machine or a host is shut down or placed into maintenance mode.
The check box /dev/hwrng source enables the cluster to use the /dev/hwrng hardwarebased random number generator instead of using the /dev/urandom device. This hardware device must be available and functioning on all hosts in the cluster if the check box is selected.
11. In addition to the General section, the New Cluster wizard includes some other sections to customize the cluster configuration:
- The Optimization section includes settings like the memory page sharing threshold for the cluster, and CPU thread handling and memory ballooning on the hosts in the cluster.
- The Migration Policy section includes settings like the VM migration policy for the cluster.
- The Scheduling Policy section supports the configuration of a scheduling policy for the cluster.
- The Console section includes the possibility to configure a custom SPICE proxy for hosts in the cluster.
- The Fencing policy section supports fencing management in the cluster.
- The MAC Address Pool supports the configuration of a custom MAC address pool for the cluster.
When done, click OK to create the new cluster.
12. A pop-up window titled Cluster – Guide Me opens. This window can be used to add hosts to the cluster immediately.
When you first create the cluster, those hosts might not be installed or configured yet. In that case, you can click the Configure Later button to complete the configuration of the cluster at a later point in time.