Jump to contentJump to page navigation: previous page [access key p]/next page [access key n]
ContentsContents
Virtualization Guide
  1. Preface
  2. I Introduction
    1. 1 Virtualization technology
    2. 2 Virtualization scenarios
    3. 3 Introduction to Xen virtualization
    4. 4 Introduction to KVM virtualization
    5. 5 Virtualization tools
    6. 6 Installation of virtualization components
    7. 7 Virtualization limits and support
  3. II Managing virtual machines with libvirt
    1. 8 Starting and stopping libvirtd
    2. 9 Preparing the VM Host Server
    3. 10 Guest installation
    4. 11 Basic VM Guest management
    5. 12 Connecting and authorizing
    6. 13 Advanced storage topics
    7. 14 Configuring virtual machines with Virtual Machine Manager
    8. 15 Configuring virtual machines with virsh
    9. 16 Managing virtual machines with Vagrant
    10. 17 Xen to KVM migration guide
  4. III Hypervisor-independent features
    1. 18 Disk cache modes
    2. 19 VM Guest clock settings
    3. 20 libguestfs
    4. 21 QEMU guest agent
    5. 22 Software TPM emulator
  5. IV Managing virtual machines with Xen
    1. 23 Setting up a virtual machine host
    2. 24 Virtual networking
    3. 25 Managing a virtualization environment
    4. 26 Block devices in Xen
    5. 27 Virtualization: configuration options and settings
    6. 28 Administrative tasks
    7. 29 XenStore: configuration database shared between domains
    8. 30 Xen as a high-availability virtualization host
    9. 31 Xen: converting a paravirtual (PV) guest into a fully virtual (FV/HVM) guest
  6. V Managing virtual machines with QEMU
    1. 32 QEMU overview
    2. 33 Setting up a KVM VM Host Server
    3. 34 Guest installation
    4. 35 Running virtual machines with qemu-system-ARCH
    5. 36 Virtual machine administration using QEMU monitor
  7. Glossary
  8. A Virtual machine drivers
  9. B Configuring GPU Pass-Through for NVIDIA cards
  10. C XM, XL toolstacks, and the libvirt framework
  11. D GNU licenses
Navigation
Applies to SUSE Linux Enterprise Server 15 SP3

30 Xen as a high-availability virtualization host

Setting up two Xen hosts as a failover system has several advantages compared to a setup where every server runs on dedicated hardware.

  • Failure of a single server does not cause major interruption of the service.

  • A single big machine is normally way cheaper than multiple smaller machines.

  • Adding new servers as needed is a trivial task.

  • The usage of the server is improved, which has positive effects on the power consumption of the system.

The setup of migration for Xen hosts is described in Section 28.3, “Migrating Xen VM Guest systems”. In the following, several typical scenarios are described.

30.1 Xen HA with remote storage

Xen can directly provide several remote block devices to the respective Xen guest systems. These include iSCSI, NPIV, and NBD. All of these may be used to do live migrations. When a storage system is already in place, first try to use the same device type you already used in the network.

If the storage system cannot be used directly but provides a possibility to offer the needed space over NFS, it is also possible to create image files on NFS. If NFS is available on all Xen host systems, this method also allows live migrations of Xen guests.

When setting up a new system, one of the main considerations is whether a dedicated storage area network should be implemented. The following possibilities are available:

Table 30.1: Xen remote storage

Method

Complexity

Comments

Ethernet

low

Note that all block device traffic goes over the same Ethernet interface as the network traffic. This may be limiting the performance of the guest.

Ethernet dedicated to storage.

medium

Running the storage traffic over a dedicated Ethernet interface may eliminate a bottleneck on the server side. However, planning your own network with your own IP address range and possibly a VLAN dedicated to storage requires numerous considerations.

NPIV

high

NPIV is a method to virtualize Fibre channel connections. This is available with adapters that support a data rate of at least 4 Gbit/s and allows the setup of complex storage systems.

Typically, a 1 Gbit/s Ethernet device can fully use a typical hard disk or storage system. When using very fast storage systems, such an Ethernet device will probably limit the speed of the system.

30.2 Xen HA with local storage

For space or budget reasons, it may be necessary to rely on storage that is local to the Xen host systems. To still maintain the possibility of live migrations, it is necessary to build block devices that are mirrored to both Xen hosts. The software that allows this is called Distributed Replicated Block Device (DRBD).

If a system that uses DRBD to mirror the block devices or files between two Xen hosts should be set up, both hosts should use the identical hardware. If one of the hosts has slower hard disks, both hosts will suffer from this limitation.

During the setup, each of the required block devices should use its own DRBD device. The setup of such a system is quite a complex task.

30.3 Xen HA and private bridges

When using several guest systems that need to communicate between each other, it is possible to do this over the regular interface. However, for security reasons it may be advisable to create a bridge that is only connected to guest systems.

In an HA environment that also should support live migrations, such a private bridge must be connected to the other Xen hosts. This is possible by using dedicated physical Ethernet devices and a dedicated network.

A different implementation method is using VLAN interfaces. In that case, all the traffic goes over the regular Ethernet interface. However, the VLAN interface does not get the regular traffic, because only the VLAN packets that are tagged for the correct VLAN are forwarded.

For more information about the setup of a VLAN interface see Section 9.1.1.4, “Using VLAN interfaces”.