ESX is currently the killer software at VMware and is declined in different version depending on what kind of support you have subscribed. However you can use the last ESXi which is pretty akin to ESX except that it is free and comes with less features and hardware support. Both ESX and ESXi can be purchased in “installable and embedded” versions. The embedded release is distributed by OEM vendors with their servers solution hence you do not have to bother with it. You can see it as a firmware settled on the hardware flash memory. It only contains the drivers it needs to work and no more. You cannot imagine make it run on an other platform than the one it is made for. The installable release comes with more hardware support and can be installed on a common hard disk or directly boot from a USB key. See http://www.vm-help.com/ and user guide for more information about that.

VMware Server (formerly GSX) is a major free software for server virtualization as type two hypervisor at VMware's. Depending on what release of VMware Server you are running you will have to focus on different things.

This section shows how to use this API to manage VMware products compliant with VMware-vix. Be sure to download and install the latest release if not installed during your VMware product installation. You also need to put every jar located in src/vmware-vix/lib and src/core/lib in the classpath. You need to set the jna.library.path variable when launching you java application (-Djna.library.path=path-to-vmwarevixlibrary). The last requirement is due to the fact that VMware-vix is a management API written in C and that we use JNA to use the C library from a Java environment. Package to use: org.ow2.proactive.virtualizing.vmwarevix.*

Here is an example showing how to connect to and manage your environment:

VMwareVMM vmm = new VMwareVMM("psychoquack.inria.fr","user","password",904,VMwareVMM.Service.vmwareServer);
VMwareVM myVM = vmm.getNewVM("/user/jmguilla/home/vmware/myUbuntu/myUbuntu.vmx");
myVM.pushData("guestinfo.myvar", "myval");
myVM.powerOn();
//some work
myVM.powerOff();
                        

The first line connects to the VMM thanks to new VMwareVMM(String,String,String,int,VMwareVMM.Server). The first tree parameters are: url of the machine running the VMware Server, a user belonging to the vmware group and his password. The following parameter is the authentication port set during the installation and the last parameter represents the type of VMM you are communicating with (in this case, vmwareServer. You can use VMware-vix to manage several VMware softwares, but we strongly recommend to use it only with VMware server prior to release 2.0). Then you either can get all registered virtual machines thanks to vmm.getVirtualMachines() or get a given virtual machine with vmm.getNewVM("/vmx/file/path"). Before powering on your vm, you can pass it some data that you will be able to read from the inside:

myVM.pushData("guestinfo.myvar","myval");

Note that to use that feature, you are compeled to have the guest tools installed within the virtual machine as explained above. To read the data within the virtual machine:

vmware-guestd --cmd "info-get guestinfo.myvar"

This is useful to define some parameters for the applications running inside the virtual machine. Finally, just power your vms on and off.

This section shows how to use this API to manage VMware products compliant with VMware-VI. Put every jar located in src/vmware-vi/lib and src/core/lib in the classpath. Package to use: org.ow2.proactive.virtualizing.vmwarevi.*

Here is an example about how to connect to and manage your environment:

VMwareVMM vmm = new VMwareVMM("https://psychoquack.inria.fr:8333/sdk","user","password");
VMwareVM myVM = vmm.getNewVM("myUbuntu");
VMwareVM myClone = myVM.clone("myUbuntuClone");
myClone.pushData("guestinfo.myvar", "myval");
myClone.powerOn();
//some work
System.out.println("myvar: " + myClone.getData("guestinfo.myvar");
myClone.powerOff();
myClone.destroy();
                        

The first line connects to the VMM thanks to new VMwareVMM(String,String,String). The tree parameters are: url (protocol aware, http://localhost:8222/sdk and https://localhost:8333/sdk by default but can be overriden during installation) of the machine running the VMware Server, a user belonging to the vmware group and his password. Then you either can get all registered virtual machines thanks to vmm.getVirtualMachines() or get a given virtual machine with vmm.getNewVM("machinesName"). Before powering on your VM, you can pass it some data that you will be able to read from the inside:

myVM.pushData("guestinfo.myvar","myval");

Note that to use that feature, you are obliged to have the guest tools installed within the virtual machine as explained above. To read the data within the virtual machine:

vmware-guestd --cmd "info-get guestinfo.myvar"

This is useful to define some parameters for the applications running inside the virtual machine. You are also able to clone the virtual machine. The clone is an immutable snapshot of the parent virtual machine and can be destroyed. Immutable means that once the vm is powered off, every changes will be lost. Furthermore, be sure to destroy a clone as this feature erases all vm components from the computer. Finally, you can power the vm on and off.

To be able to manage Microsoft Hyper-V with this implementation, you must put every jars located in src/hyperv-winrm/lib and src/core/lib in the class path. This module uses Wiseman to leverage WS-Management profile. Ensure that the user you use to connect to remote Hyper-V server owns sufficient permissions to manage Hyper-V.

All network communications are made over HTTP or HTTPS. With Windows Server 2008 releases, the WS-Management implementation is installed by default, you can find it in WINROOT/system32/winrm.exe. To be able to use it, you just have to activate it. You can use the folowing tutorials:

If you decide to use HTTPS, you may have to set some variables on the Java client side and to add the server's public key to one of your keystores:

You may have noticed that this solution has a real advantage over the WBEM implementation, it is firewall friendly...

Here is an example showing how to connect to and manage your environment:

System.setProperty("javax.net.ssl.keyStore", "/home/user/.keystore");
System.setProperty("javax.net.ssl.trustStore", "/home/user/.keystore");
System.setProperty("javax.net.ssl.trustStorePassword", "keyStorePassword");
System.setProperty("javax.net.ssl.keyStorePassword", "keyStorePassword");
HyperVVMM vmm = new HyperVVMM("https://hyperv.address.com/wsman", "hypervUser", "hypervUserPassword");
HyperVVM vm = vmm.getNewVM("Ubuntu");
HyperVVM clone = vm.clone("UbuntuClone");
clone.pushData("testVar", "testVal");
clone.powerOn();
//some work
System.out.println("test: " + clone.getData("testVar"));
clone.powerOff();
clone.destroy();
                        

The four first lines allows HTTPS communication, the fifth line connects to the VMM thanks to new HyperVVMM(String,String,String). The tree parameters are: url, user with enough permissions and his password. For the url, don't forget the protocol and the trailing /wsman... Then you can either get all registered virtual machines thanks to vmm.getVirtualMachines() or get a given virtual machine with vmm.getNewVM("machinesName"). Before powering on your vm, you can pass it some data that you will be able to read from the inside:

myVM.pushData("myvar","myval");

To get the information from the inside of the virtual machine, if guest tools are installed, read the registry key: HKLM/SOFTWARE/Microsoft/Virtual Machine/External. You also can use the method HyperVVM#getKvpExchangeData("testVar") to read this value. To set a variable from the inside of the virtual machine, if guest tools are installed, refer to: this blog entry. and read it thanks to HyperVVM#getGuestExchangeItem("testVar"). If guest tools are not installed, use HyperVVM#pushData and HyperVVM#getData. pushData first tries to delete a previously set kvp with the same key and then call pushKvpExchangeData. getData first tries to getGuestExchangeItems, if this call fails, then tries getGuestExchangeItem.

To be able to manage Microsoft Hyper-V with this implementation, you must put every jars located in src/hyperv-wmi/lib and src/core/lib in the class path. This module uses J-Interop to use DCOM communication protocol. Ensure that the user you use to connect to remote Hyper-V server owns sufficient permissions to manage Hyper-V.

To manage this tool, you must allow WMI/MSRPC communication in your firewall configuration. Administrative tools -> Component Services -> right click on My Computer Edit limits & default for Acces Permissions and Launch and Activation Permissions for the user you will use to connect to virtualization service. For firewall rules and remote access: http://www.windowsreference.com/windows-server-2008/install-configure-hyper-v-in-windows-server-2008-64-bit-core-install/ With Hyper-V running on a Microsoft Server R2 edition you also must edit registry. Launch regedit.exe, select the KEY HKEY_CLASSES_ROOT//CLSID//{76A64158-CB41-11D1-8B02-00600806D9B6} (ie. WBEM Scripting Locator) The owner of this key is Trusted Installer which is a service with special permissions. To be able to use advanced scripting features using j-interop, we need full permissions for this KEY. The way to get that done is straight forward, we must change the owner of the key (right click on the directory, advanced, owner...) and allow full access to the new owner. An other alternatives may be available, see:

Here is an example showing how to connect to and manage your environment:

HyperVVMM vmm = new HyperVVMM("hyperv.address.com", "user", "pwd");
HyperVVM vm = vmm.getNewVM("myUbuntu");
HyperVVM clone = vm.clone("myUbuntuClone");
clone.pushData("testVar", "testVal");
clone.powerOn();
//some work
System.out.println("test: " + clone.getData("testVar"));
clone.powerOff();
clone.destroy();
                        

The first line connects to the VMM thanks to new HyperVVMM(String,String,String). The tree parameters are: url, user with enough permissions and his password. Then you can either get all registered virtual machines thanks to vmm.getVirtualMachines() or get a given virtual machine with vmm.getNewVM("machinesName"). Before powering on your vm, you can pass it some data that you will be able to read from the inside:

myVM.pushData("myvar","myval");

To get the information from the inside of the virtual machine, if guest tools are installed, read the registry key: HKLM/SOFTWARE/Microsoft/Virtual Machine/External. You also can use the method HyperVVM#getKvpExchangeData("testVar") to read this value. To set a variable from the inside of the virtual machine, if guest tools are installed, refer to: this blog entry. and read it thanks to HyperVVM#getGuestExchangeItem("testVar"). If guest tools are not installed, use HyperVVM#pushData and HyperVVM#getData.

Libvirt (see http://libvirt.org/) is a management API project for virtualized infrastructures. Its aim is to provide an unique API / Interface to manage several VMM. You can notice that it currently supports: Xen (on Linux and Solaris), QEMU, KVM, LXC, OpenVZ, User Mode Linux, VirtualBox, and that other supports are announced (VMware server and ESX).

If you used Virt-manager to install and set up your virtualized environment, you must already have Libvirt installed on your computer and your virtual machines registered. Otherwise, download the project here and install it. Then, you have to register your previously created virtual machines. You can read the associated documentation to get it done.

When evaluating this environment, we noticed different "hot points" about which one you have to be careful. Libvirt may use different mates for authentication, just be sure that you have enable the good user with sufficient permissions and using the good authentication mechanism. Be sure to understand the basic knowledge about URLs. Understand that, when using Xen Oss, you will have to perform extra configuration on the server side to set up your Xen VMM the good way (to use either sockets, http, legacy remote control ...).

Here is an example showing how to connect to and manage your environment:

LibvirtVMM vmm = new LibvirtVMM("qemu+ssh://psychoquack.inria.fr/system");
LibvirtVM myVM = vmm.getNewVM("machinesName");
myVM.powerOn();
//some work
myVM.powerOff();
                    

The first line connects to the VMM thanks to new LibvirtVMM(String). The String parameter is the url (see URL) used to connect to the Libvirt daemon. Then you can either get all registered virtual machines thanks to vmm.getVirtualMachines() or get a given virtual machine with vmm.getNewVM("machinesName"). With that implementation, you can only power the vm on and off and cannot pass information to it.

Xen OSS is the base component of XenServer. It is a type one hypervisor which comes as a particular kernel for several linux distribution (redhat, fedora, ubuntu, ...). The best way to have a fully personalized Xen is to compile sources and build its own kernel. If you want to use that software: Download the source tarball, CAREFULY read the readme (you also can use that tutorial) and build your own kernel(s):

cd /usr/src
wget http://bits.xensource.com/oss-xen/release/3.3.1/xen-3.3.1.tar.gz
tar xvf xen-3.3.1.tar.gz
ln -s xen-3.3.1 xen
cd xen

Now we compile Xen with 2 targets kernels

make world vmxassist=y KERNELS="linux-2.6-xen0 linux-2.6-xenU"
install.sh

The vmxassist=y flag means that we want to allow HVM (Hardware-enhanced Virtual Machine, requires hardware suppport) guests to run, so we ask Xen to build a specific container that will be available in /usr/lib/xen/boot/hvmloader. To be able to compile such a Kernel, you must ensure that the packages glibc-devel, dev86, libvncserver,SDL and SDL-devel are installed. If you want, you can make them more efficient...

cd /usr/src/xen/build-linux-2.6.18.8-xen0_[arch]
make menuconfig

Check the following options:

Xen --> [*] Privileged Guest (domain 0)
File systems --> [*] Quota support
			     [M] Old quota format support
			     [M] Quota format v2 support

Device Drivers ---> Networking support --> [M] Dummy net driver support

Networking support --> Networking options --> [*] Network packet filtering (replaces ipchains) --> IP: Netfilter Configuration --> [M] IP tables support (required for filtering/masq/NAT).
						

then:

make
make modules
make modules_intall
make install
						

Just repeat the whole procedure unchecking

Xen --> [ ] Privileged Guest (domain 0)
						

for the domU kernel. Note that you can forget that last build as the last releases of Xen OSS allows to run both host and guest with the same kernel config and the same efficiency. Just be sure that the modified /boot/grub/menu.lst is up to date and reboot on your newly created dom0 install.

Once your Xen Oss virtual environment is installed, you have to setup your virtual machines inside. To help you, you have several client software, among those one: Convirt (aka Xenman) and virt-manager. Our favorite one for the installation step is Convirt. It will allow you to install both paravirtualized and hardware assisted virtualized guest operating systems (the setup "by hand is feasible" for paravirtualized but more complicated for others and deals with Xen configuration files, thus we encourage to avoid it...). Here is one of the good tutorial you can find on the web. Virt-manager is better for virtual machine management and handles more virtual environments (furthermore it is built on top of libvirt API).

That hypervisors are all based on Qemu. They are all type two hypervisors that allow emulation (Qemu) and hardware assisted virtualization(Qemu-KVM and KVM). The best advantage of using that softwares (in fact, only Qemu is concerned) is that it is possible to use them entirely in user space (do not need root account for installation and use). If you want to use that solution, download one of the products (Qemu-KVM is an add-on for Qemu): Qemu or KVM and install it thanks to the README explanations.

To install virtual machines and manage your environment you can use Convirt (aka Xenman) or virt-manager or try to get it done by hand.

dd if=/dev/zero of=disk.img bs=512k count=1 seek=2000
mkfs.ext2 disk.img
					

build a raw disk image with a capacity of 1Go and an ext2 file system (for linux based systems).

kvm -net nic -net user -m 512 -boot d -cdrom /home/jmguilla/isos/fedora.iso disk.img
					

install your OS thanks to the .iso file

kvm -net nic -net user -m 512 -boot c disk.img
					

to boot your virtual machine once the operating system is installed.