Making virtualisation play nice with application delivery

By Jim Metzler
Thursday, 02 July, 2009


Virtualisation is touted as central to helping IT organisations optimize and secure application performance in a cost effective manner, but it can also present its share of application delivery challenges.

In this discussion, we examine the impact of virtualisation on application delivery and performance.

The term virtualisation is used in a variety of ways to describe a number of techniques. Two of the most common uses of the term virtualisation involve a logical abstraction of physical systems that allows one of the following:

  • A single physical system to be partitioned to appear as multiple independent logical systems; e.g., multiple VLANs defined on a single physical LAN or multiple VPNs on a single WAN link.
  • Multiple physical systems to appear as a single logical system; e.g., a compute cluster with a single system image. When a host is added to a cluster, the host's resources become part of the cluster's resources, and the cluster manages the resources of all hosts within it.

One observation that can be drawn from the preceding definition is that virtualisation is not a new concept, because VLANs and VPNs have been widely deployed for well over a decade.

Most of the current interest in virtualisation revolves around virtual servers in part because virtualizing servers can result in significant cost savings. The phrase virtual machine (VM) refers to a software computer that, like a physical computer, runs an operating system and applications. An operating system on a virtual machine is called a guest operating system. The guest OSs can be instances of a single version of one OS, different releases of the same OS, or completely different OSs; e.g., Linux, Windows, Mac OS-X or Solaris. A thin software layer called a virtual machine monitor (VMM) or hypervisor creates and controls the virtual machine's other virtual subsystems.

Figure 1 shows how Ethernet Networking I/O is typically virtualised by VMM software. The VMs within a virtualised server tend to share a conventional physical Ethernet NIC (PNIC) to connect to a data center LAN. The VMM provides each VM with a virtual NIC (VNIC) and creates a virtual network to provide the connectivity between the VNICs and the PNIC. This virtual network is based on a virtual switch that is often referred to as a vSwitch.

Software-based virtual Ethernet IO
Figure 1: Software-based Virtual Ethernet I/O

In the traditional environment consisting of physical servers connected by a physical switch, IT organisations can get detailed management information about the traffic that goes between the servers from the physical switch. Unfortunately, that same level of management information is typically not provided by a vSwitch. The resultant lack of visibility into the traffic flows between and among the VMs on the same physical platform affects security, performance monitoring and troubleshooting.

The potential also exists for IT organisations to combine too many VMs onto a physical server. This can result in performance problems caused by factors such as limited CPU cycles or I/O bottlenecks that result from oversubscribed physical NICs. These problems can occur in a traditional physical server, but they are more likely to occur in a virtualised server because of the consolidation onto a single physical server of multiple VMs, all of them contending for scarce resources. As a result, management tasks such as performance management and capacity planning are more important in a virtualised environment than they are in a physical environment. This means that IT organisations must be able to continuously monitor in real time the utilisation of both physical servers and VMs. This capability allows IT organisations to avoid both over- and underutilisation of server resources such as CPU and memory and to allocate and reallocate resources based on changing business requirements. This capability also enables IT organisations to implement policy-based remediation that helps the organisation to ensure that service levels are being met.

Another challenge created by the deployment of virtual servers is that IT organisations now have to manage VM sprawl. The phrase VM sprawl refers to the proliferation of VMs combined with the IT organisation's lack of visibility and control over the VMs. In addition, most vendors of server virtualisation provide the ability to move VMs quickly and easily from one physical server to another. This ability is a classic good news/bad news situation. The good news is that this ability contributes to resource agility, high availability and energy efficiency. The bad news is that it can be challenging to ensure that the migrated VM retains the same security, storage access, and QoS configurations and policies. Keeping all the required configurations synchronized requires linkages among the management systems for physical and virtual servers, network devices and storage.

There is no doubt that virtualisation creates some application delivery challenges. There is also no doubt that virtualisation helps IT organisations deploy techniques to optimize and secure application performance in a cost-effective manner. For example, just as devices such as servers can be virtualised, so can appliances such as WAN optimisation controllers (WOCs). A virtual appliance is based on network appliance software, together with its operating system, running in a virtual machine in a virtualised server. Virtual appliances can include WOCs, application delivery controllers (ADCs), firewalls, and performance monitoring solutions, among others. A virtual appliance offers the potential to alleviate some of the management burdens in branch offices because most of the provisioning, software updates, configuration and other management tasks can be automated and centralised at the data center. Of course, virtualised appliances create their own set of challenges. For example, just because the software is capable of running in a VM does not necessarily mean that it performs well in that environment.

This discussion of the emerging applications delivery challenges will be continued in ashtonmetzler.blogspot.com/.

Related Articles

Private AI models: redefining data privacy and customisation

Private AI signifies a critical step forward towards a more secure, personalised and efficient...

Why having an observability strategy is critical for effective AI adoption

As organisations continue to adopt AI and put it to work in a variety of innovative ways, many...

What you need to know to build a winning AI strategy

For organisations that have yet to start investing in AI solutions, it's not too late to use...


  • All content Copyright © 2024 Westwick-Farrow Pty Ltd