Inter-VLAN Routing is the capability to route network traffic between VLANs. This feature enables you to set up Virtual Private Clouds (VPC) that can hold multi-tier applications. These tiers are deployed on different VLANs that can communicate with each other. You can provision VLANs to the tiers your create, and VMs can be deployed on different tiers, such as Web, Application, or Database. The VLANs are connected to a virtual router, which facilitates communication between the VMs. In effect, you can segment VMs by means of VLANs into different networks that can host multi-tier applications. Such segmentation by means of VLANs logically separate application VMs for higher security and lower broadcasts, while remaining physically connected to the same device.

This feature is supported on XenServer and VMware hypervisors.

A Site-to-Site VPN connection helps you establish a secure connection from an enterprise datacenter to the cloud infrastructure. This allows users to access the guest VMs by establishing a VPN connection to the virtual router of the account from a device in the datacenter of the enterprise. Having this facility eliminates the need to establish VPN connections to individual VMs.

The supported endpoints on the remote datacenters are:

You can now create data volumes on local storage. The data volume is placed on the same XenServer host as the VM instance that is attached to the data volume. These local data volumes can be attached to virtual machines, detached, re-attached, and deleted just as with the other types of data volume. In earlier releases of CloudStack, only the root disk could be placed in local storage.

Local storage is ideal for scenarios where persistence of data volumes and HA is not required. Some of the benefits include reduced disk I/O latency and cost reduction from using inexpensive local disks.

In order for local volumes to be used, the feature must be enabled for the zone.

You can create a data disk offering for local storage. When a user creates a new VM, they can select this disk offering in order to cause the data disk volume to be placed in local storage.

You can not migrate a VM that has a volume in local storage to a different host, nor migrate the volume itself away to a different host. If you want to put a host into maintenance mode, you must first stop any VMs with local data volumes on that host.

Local storage support for volumes is available for XenServer, KVM, and VMware hypervisors.

A tag is a key-value pair that stores metadata about a resource in the cloud. Tags are useful for categorizing resources. For example, you can tag a user VM with a value that indicates the user's city of residence. In this case, the key would be "city" and the value might be "Toronto" or "Tokyo." You can then request CloudStack to find all resources that have a given tag; for example, VMs for users in a given city.

You can tag a user virtual machine, volume, snapshot, guest network, template, ISO, firewall rule, port forwarding rule, public IP address, security group, load balancer rule, project, VPC, network ACL, or static route. You can not tag a remote access VPN.

You can work with tags through the UI or through the new API commands createTags, deleteTags, and listTags. You can define multiple tags for each resource. There is no limit on the number of tags you can define. Each tag can be up to 255 characters long. Users can define tags on the resources they own, and administrators can define tags on any resources in the cloud.

A new optional input parameter, "tags," has been added to many of the list* API commands. The following example shows how to use this new parameter to find all the volumes having tag region=canada OR tag city=Toronto:

command=listVolumes
&listAll=true
&tags[0].key=region
&tags[0].value=canada
&tags[1].key=city
&tags[1].value=Toronto

The following API commands have the new "tags" input parameter:

Some changes have been made to the Amazon Web Services API compatibility support in order to accommodate the new tagging feature.

New APIs:

Changed APIs:

With the addition of Secure Console feature, users can now securely access the VM consoles on the XenServer hypervisor. You can either SSH or use the View Console option in the Management Server to securely connect to the VMs on the XenServer host. The Management Server uses the xapi API to stream the VM consoles. However, there is no change in the way you can access the console of a VM. This feature is supported on XenServer 5.6 and 6.0 versions.

This release supports creating VMs without starting them on the backend. You can determine whether the VM needs to be started as part of the VM deployment. A VM can be deployed in two ways: create and start a VM (the default method); create a VM and leave it in the stopped state.

A new request parameter, startVM, is introduced in the deployVm API to support the stopped VM feature. The possible values are:

Existing data can now be made accessible to a virtual machine. This is called uploading a volume to the VM. For example, this is useful to upload data from a local file system and attach it to a VM. Root administrators, domain administrators, and end users can all upload existing volumes to VMs. The upload is performed by using HTTP. The uploaded volume is placed in the zone's secondary storage.

This functionality is supported for the following hypervisors:

One or more hosts can now be designated for use only by high-availability (HA) enabled VMs that are restarted due to a host failure. Setting up a pool of such dedicated HA hosts as the recovery destination for all HA-enabled VMs make it easier to determine which VMs are restarted as part of the high-availability function. You can designate a host as a dedicated-HA restart node only if the Dedicated HA Hosts feature is enabled by setting the appropriate global configuration parameter.

This release supports Amazon Web Services APIs, including Elastic Compute Cloud (EC2) API. Fidelity with the EC2 API and the installation experience for this functionality are both enhanced. In prior releases, users were required to install a separate component called CloudBridge, in addition to installing the Management Server. For new installations of CloudStack 4.0.0-incubating, this software is installed automatically along with CloudStack and runs in a more closely integrated fashion. The feature is disabled by default, but can be easily enabled by setting the appropriate global configuration parameter and performing a few setup steps.

The Nicira NVP plug-in allows CloudStack to use the Nicira solution for virtualized network as a provider for CloudStack networks and services. In CloudStack 4.0.0-incubating this plug-in supports the Connectivity service. This service is responsible for creating Layer 2 networks supporting the networks created by guests. When a tenant creates a new network, instead of a traditional VLAN, a logical network will be created by sending the appropriate calls to the Nicira NVP Controller. The plug-in has been tested with Nicira NVP versions 2.1.0, 2.2.0 and 2.2.1.

CloudStack 4.0.0-incubating supports using a CAStor cluster as the back-end storage system for a CloudStack S3 front-end. The CAStor back-end storage for CloudStack extends the existing storage classes and allows the storage configuration attribute to point to a CAStor cluster. This feature makes use of the CloudStack server's local disk to spool files before writing them to CAStor when handling the PUT operations. However, a file must be successfully written into the CAStor cluster prior to the return of a success code to the S3 client to ensure that the transaction outcome is correctly reported.

The S3 multipart file upload is not supported in this release. You are prompted with proper error message if a multipart upload is attempted.

This release adds Clustered Logical Volume Manager (CLVM) storage support for KVM hosts. With this support, you can use CLVM as primary storage.

The CLVM support for KVM allows root and data disks (primary storage) to reside on Linux logical volumes. The administrators are required to configure CLVM on the KVM hosts independent of CloudStack. When the volume groups are available, an administrator can simply add primary storage of type CLVM, providing the volume group name. Then CloudStack creates and manages logical volumes as needed.

CLVM also supports Snapshots. CloudStack creates an LVM snapshot, copy the applicable logical volume to the secondary storage in the qcow2 format, and then delete the LVM snapshot.

You can now use Rados Block Device (RBD) to run instances on Apache CloudStack 4.0.0-incubating. This can be done by adding a RBD pool as primary storage. Before using RBD, ensure that Qemu is compiled with RBD enabled, and the libvirt version is at least 0.10 with RBD enabled on the KVM host

Create a disk offering for RBD so that you can ensure that StoragePoolAllocator chooses the RBD pool to deploy instances.