VMworld 2015: VMware touts new technologies for managing containers and live migration of virtual machines
- 01 September, 2015 00:06
VMware executives demonstrated new virtualization technologies designed to run in the data centers of the future at the company's annual VMworld conference.
VMware showed off a new capability of moving a live, running virtual machine (VM) from one data center in California to another in Virginia. Executives also demonstrated how the company is working with containers, a rapidly emerging virtualization technology popularized by Docker.
The ability to move live VMs is an advance on the company's vMotion technology, which has been available for several years. The vMotion technology offers a way to move a VM from one part of the data center to another, while keeping the application inside the VM running.
Now that migration can be done across different data centers, something that has not been possible to do before with VMs.
"It's the same vMotion that we have all been familiar with, but now we are able to accelerate, compress and move the VMs live from one location and one domain to another location and domain," said Raghu Raghuram, VMware's executive vice president of cloud infrastructure and management.
Traditionally, moving a live VM from one data center network to another has been impossible, given that each network operates under a different network subnet. To facilitate the technology, VMware was able to take advantage of its network virtualization software (NSX) which makes it possible for the VM to retain its network address even as it moves to another physical network.
Such a capability, available as a technology preview, will make it even easier for organizations to bridge their internal cloud networks with public cloud capabilities, which can be handy for purposes of disaster recovery, application development, and for handling bursts of application activity.
The company showed off some advances in the realm of containers as well.
Containers are lighter than the full-sized virtual machines, such as VMware's, because they are not packaged with an entire operating system. Instead, they borrow capabilities from the host operating system, which means they can be started much more quickly, as well as moved around more easily.
"Containers come up again and again in conversations," said Ray O'Farrell, VMware's chief technology and development officer.
With VMware offering a complete stack of software for virtualizing the data center, the company had to address the growing popularity of containers. Typically, VMware's vSphere software does not recognize containers, only VMs, which makes it difficult for the administrator to manage containers, and introduces security risks.
"There is an opportunity to have greater management when it comes to managing containers in our current infrastructure," O'Farrell said.
To answer this challenge, VMware has developed vSphere Integrated Containers, available as a technical preview, which allows the vSphere management software to work with containers.
"Containers will be first-class citizens on vSphere. You can manage both traditional applications inside VMs, and next-generation applications inside of containers, side by side, fully consistently on one platform," said Kit Colbert, the company's chief technology officer for cloud-native apps.
To support containers, the company reengineered the VM, moving some of its functionality, such as the container engine, to the vSphere software itself, in effect, making vSphere Integrated Container work as a container, while still keeping the management characteristics of the VM.
"The container is the VM and the VM is the container and that gives us all these great management capabilities," Colbert said.
The company has also released, as a preview, software for managing large fleets of containers, called the VMware Photon Platform. Photon is being designed for organizations that are building cloud-native applications. It features an API (application programming interface) and can run any one of a number of container orchestration tools, such as Kubernetes or Mesophere. The core Photon engine has been released as open source.