Network approach optimises the business benefits of data center virtualisation

By Charles Ferland, VP of EMEA, BLADE Network Technologies

With today’s limited budgets, data centre administrators are looking for ways to increase hardware utilization, power savings, availability, and scalability. Vitalising individual servers can accomplish all of these things, yet those advantages represent just a fraction of virtualisation’s potential. Expanding virtualisation to the network lets virtual machines (VMs) move between physical servers while applications run, providing capacity when and where it is needed without adding expensive and underutilised platforms. What’s more, dynamic VM creation and movement lets administrators respond quickly to new requests, increasing user productivity and customer satisfaction.

Until recently, many data centre administrators hesitated to take advantage of mobile VMs because of the time demands of managing the related networking requirements. Now automated solutions can safely create and move virtual machines, freeing network administrators from those responsibilities and bringing the full potential of virtualisation technology to even resource-strapped data centres.

Expanding Virtualisation Benefits to the Network
It’s morning on a typical work day. Employees are logging in to their systems and customers are accessing web sites. Very quickly, the demand for server resources grows from minimal to extensive. Using network virtualisation, VMs that have been consolidated on a few servers overnight can migrate to additional machines that are newly powered up to meet capacity requirements. Users are unaware of this movement; they simply take advantage of a high-performing environment with ready access to business applications located throughout the data centre. At the end of the day, when demand is lower, the VMs migrate back to a few servers, easing management and reducing power costs.

Daily VM migration is just one practical application of mobile VM technology. Moving VMs also makes sense at the end of a fiscal quarter, when financial applications require more processing power, or when a business launches a new service that generates extensive web site activity. VMs can also be moved to support business continuity when servers are taken off-line for maintenance or in the case of a disaster. The examples are almost limitless, and the benefits are enormous.

To take advantage of these benefits, administrators must perform some network-related activities. The network management requirements associated with moving VMs between hardware platforms are not tremendously complex, but they are critically important and can be time-consuming.

As VMs move around the data centre, the network-level policies associated with each VM must move with them. These policies govern factors such as security and Quality of Service (QoS), and differ based on the users and applications.

Maintaining appropriate network policies for every business application is critical to business operations. Finance and Human Resource departments must have very strict security policies to protect sensitive corporate and personnel data. Research and development teams are extremely concerned with protecting valuable intellectual capital, and require a highly-secure application environment. For applications such as voice and video, performance is likely to be a greater concern than security, so the QoS configuration must be optimised.

Before a VM is moved from one physical server to another, its associated networking port configuration and security policies must be correctly set for the destination server. This task must be accomplished accurately. Incorrectly moving security policies could prevent users from accessing critical applications, open a business to attack by making sensitive data available to the wrong users, or even cut off network access for VM users altogether.

Manually configuring a destination port for each VM can be time-consuming. If a data centre has a lot of servers and a lot of VM movement, the time required to configure ports and oversee policy movement could be extensive. In addition, network administrators must coordinate with server administrators as they perform these tasks, which can create a complex workflow challenge. If the data centre is very active and it makes sense to move VMs more than once per day, the effort to manage the migration becomes impractical relative to the value obtained.

The solution to taking advantage of network virtualisation without overwhelming administrators is automating the network management tasks associated with VM movement.

Automating Network Management for VM Movement
The key to automating the network aspects of VM movement is creating a network switch with the intelligence to differentiate between individual VMs within a physical server. A traditional network switch views a server through the physical port to which it is connected. Through that port, the switch is aware of one set of configuration information. A switch containing intelligent software that is aware of the individual VMs within the server and the network traffic related to each VM can extend the individual VM attributes into the network. As the VM migrates through the network, the switch can track that migration and ensure that the network settings migrate along with the VM.

An important aspect of VM-aware switch software is the ability to work with any hypervisor. Also known as a virtual machine monitor (VMM), the hypervisor is a critical piece of virtualisation software that allows multiple operating systems to run on a single host platform. Hypervisors are available from software vendors such as Citrix, Microsoft and VMware, and from some open source projects such as Xen.

The ability to work with multiple hypervisors is crucial because data centres commonly run different hypervisors. A research and development department might prefer one hypervisor because it gives users more control over their servers, while data administrators might standardise on another hypervisor from a particular vendor for other users. Or, data centre administrators might choose to gradually migrate to a different vendor’s hypervisor for pricing or other reasons, creating a temporarily mixed environment.

Other valuable attributes of a switch-based virtualisation product include:
• The ability to track VM movement without altering Ethernet packets to minimise overhead.
• Intelligence built into the switch that relieves network administrators and the server central processing unit (CPU) of any responsibility for tracking VMs and VM policy movement.
• An architecture that lets users readily configure network settings at the VM level, increasing productivity.
• The ability to operate in a multi-vendor network environment to maximise flexibility.
• An interface with management platforms to ease administration.
• A robust implementation that can operate at the data centre and cloud scale with minimal administrative burden and oversight, increasing availability.

Illustration 1.  BLADE Network Technologies SmartConnect™ Switch

Automated Solutions for Mobile VMs
Solutions that automate the network aspects of virtualisation are now available. The first network product to automatically support VM migration is SmartConnect™ with VMready™ from BLADE Network Technologies. VMready software allows the network switch to discover VMs and correlate performance and security settings to a virtual port instead of a physical port. An administrator using VMready has the granularity to see each VM with its associated virtual ports. VMready also monitors the creation and movement of virtual machines, and makes sure that network settings are properly migrated along with the VM. These actions help reduce the risk of an application outage due to a misconfigured network as well as the security risk of exposing a sensitive application to unauthorised users.

Illustration2.  Benefits of a VMready™ Network Switch 

A Dynamic Data Centre
New technology that automates the labour-intensive process of ensuring that the appropriate network polices are in place before a VM moves to a new host makes dynamic VM migration practical. Now administrators can support automatic scale-up of application capacity to match changing business requirements, and dynamic load rebalancing that optimises server utilisation across the data centre. A network-virtualised data centre becomes a cloud-ready infrastructure that supports cost-effective change and growth, and is well-positioned to quickly take advantage of lucrative new business opportunities.

Illustration3.  Virtualisation Ready Network Solution