Multivendor UC & the cloud challenge
While unified communications has been around for a while, how can it work in multivendor environments where cloud services are still new? How do we move from bare metal systems to cloud and what does that mean for the user experience?
Virtualisation has already transformed the way business applications are deployed in data centres and unified communications managers are no different in wanting to transform the way they deploy and deliver UC. Burdened with the challenge of supporting a medley of servers, operating systems and applications acquired during numerous purchasing cycles, many organisations are looking to simplify and optimise their IT operations. One of the ways they can achieve this is through server virtualisation.
But will this model work for unified communications? Introducing virtualisation raises the bar on performance and accentuates the challenges of delivering quality VoIP and UC services because of the need for real-time host and guest performance. Until quite recently, real-time applications in virtualised production environments weren’t an ideal combination, but with maturing technologies and major vendor support, virtualised UC is now a reality.
As UC moves into the cloud where clients pay for services only as needed, customers will benefit from having access to a set of remote resources and can avoid purchasing, installing and maintaining hardware that depreciates and is often significantly underutilised.
A truly integrated management solution providing high-level and deep drill-down metrics into the performance, availability, capacity and quality of the host, guests and applications will enable service delivery insight for private and public clouds alike.
Hardware implies permanence and inflexibility. Metal a little more malleability. Software has the ability to transform both. You can put an entirely new program in the hardware and create an entirely new experience for the user. This is the way it is with virtualisation.
But with this transformation, the physical elements increase in criticality because they’re carrying a greater load and more will be asked of them. The right decisions need to be made quickly because more people or processes are affected. If mean times to identify, convince and repair are not rapid, the overall quality of the user experience will suffer and service levels can be breached.
When hardware is virtualised, with multiple guests acting as individual servers it’s critical to know that it’s up to the job. When a guest running a continuity-critical application makes a request in real time it is without regard for other host activity. Hence, both guests and hosts can come under performance pressure. With this type of environment problems can exist in any one of the layers. It can be within the physical hardware or the virtual machines as well as the applications themselves. Because of this, problem detection needs to be multilayer, multivendor and multitechnology. A rich collection of UC and virtualisation management metrics for the host, guests and UC applications will give you a high-level view as well as the granular detail you need to monitor and troubleshoot individual components.
A low-level virtual machine manager, known as a hypervisor, enables the server hardware - the metal - to function as a host. Once the hypervisor is installed, its only task is to run guest operating systems, and accept and arbitrate resource requests for guest VMs.
To view the impact of applications and guests on host performance, you need to monitor key host performance metrics like CPU utilisation, memory consumption, and disk and network usage. Problem identification, analysis and resolution can be highly complex so comparing host metrics side by side is invaluable in being able to identify and address any resource contention.
In this way you’ll be able to easily recognise if a particular VM is taking too many of the host’s resources and affecting other VMs’ performance. You’ll be able to rapidly isolate the component at fault, and avoid finger pointing between different support teams.
As well as troubleshooting, monitoring usage in real time helps deliver the economic benefits of reduced purchasing, installation and maintenance of frequently underused hardware. Capacity planning reporting allows administrators to consolidate and optimise existing servers and increase virtualisation density, while ensuring there is room for growth. Private and public cloud providers alike can better predict monthly operating expenses rather than purchasing and deploying additional capital assets.
However, these benefits must be carefully managed to ensure that cost rationalisation does not impact the real-time requirements of VoIP and UC. As such, it’s vital to have access to flexible visualisations that reflect the business impact of the applications should they experience latency or fail.
The consolidation of multiple applications onto fewer servers may lower hardware provisioning, configuration and maintenance costs, but it can also increase the risk of hardware failure. Sharing hardware resources between applications running in guest machines creates a need to monitor resource contention between guests and across CPUs, memory and disk interfaces.
You’ll want to see the top guests by CPU, disk use and memory consumption as well as the amount of time each guest waited for physical CPU cycles. Clicking on a flat line or spike to identify a particular guest will mean you can select it rapidly without scrolling through a list of hundreds or thousands of guests. Once you’ve selected the guest, drill-down capabilities will help identify resource contention that can cause performance bottlenecks and affect the value of UC to your users.
Finally, it’s a distinct benefit if you have a choice of combining vCenter Server metrics with more granular information collected directly from each guest. You’ll then be able to accurately monitor the host’s performance and view deep drill-down process and application-level information on a guest-by-guest and host-by-host basis.
The applications that can be found in a UC solution often need to be highly aware of each other. In direct contrast, the VMs they run in must not. Fortunately, the deliberate convergence of communications technologies can coexist with the equally deliberate separation of virtual machines. That is not to say that the applications themselves need not be aware of the other servers, but there must be no blurring or overlap between the virtual servers themselves.
Therefore, UC applications’ performance management requires a blend of insight to both physical and virtual environments. It should be able to leverage traditional collection methods as well as state-of-the-art web services and SIP events to provide voice quality, device, performance and availability statistics.
It should also enable you to correlate VoIP quality with network and virtualisation performance. This helps you identify bottlenecks and control communication costs by meeting and optimising UC requirements as well as reducing server costs through virtualisation.
An important aspect to remember when managing applications is the difference between quality of service (QoS) and the user’s quality of experience (QoE). From a technical standpoint you need to manage the underlying infrastructure and service delivery.
What matters to the user, however, is that they can enhance their productivity, save time and be more efficient through using unified communications.
As a customer you’ll expect to benefit from cloud computing by having access to a set of remote resources and avoid purchasing, installing and maintaining hardware that depreciates and is often significantly underused. It also eliminates the need to hire, train and manage a skilled 24/7 IT staff. Changes that took days or weeks with physical servers can be done in minutes. In this way you can predict monthly operating expenses rather than depleting and underutilising valuable capital resources.
You can store large amounts of data from anywhere without worrying about maintenance. Cloud computing service level agreements should also ensure there is no change in performance during peak times and allows your business to have on-demand servers available in minutes, with dynamic scaling, ensuring that you only pay for what you use.
If you choose to implement your own virtualised UC ecosystem, you’ll need to manage a complex environment with multiple vendors, technologies and quite possibly versions selected over many purchasing cycles. It’s likely to include a variety of servers, operating systems and applications running in virtual machines. Because of this you’ll benefit from much the same management approach as a service provider.
When compared to the public cloud this is a more expensive solution but provides more security and privacy, and you’ll have complete control of your data.
A single pane of glass monitoring for VoIP, email infrastructure, UC, cross-platform and virtual servers in real time will help you meet and manage service level agreements. You’ll be proactively alerted to status changes in UC infrastructure performance, so you can plan and execute timely remedial action.
In our Tech Insights series, we speak with ICT leaders to get their view on what the year ahead...
As more and more enterprises finally ditch the PABX for the latest in unified comms, the benefits...
There are three key reasons why organisations with a contact centre should look to implement a...