Heading into the cloud: Managing network performance across a unified fabric
This vendor-written tech primer has been edited by Network World to eliminate product promotion, but readers
should note it will likely favor the submitter's approach.
The adoption of virtualization and
cloud-based services is accelerating, but a
byproduct enterprises may not have considered is that it has never been more challenging to gain accurate and
performance visibility in a uniform way.
Existing virtualization tools are either not available to the operations team or lack the standard operations
center functionality required to proactively assure application performance and reliability. And conventional performance
management solutions designed to collect specific types of data that must be aggregated and reported on
centrally, can't scale or provide immediate visibility across the enterprise's network infrastructure.
TECH DEBATE: What to look for in network
These limitations are compounded by the fact that compute, network and storage resources are shared when
infrastructure is being provided as a service, making it difficult for an enterprise to assess which of its applications are being supported, how they are
performing and whether they are receiving the necessary bandwidth.
Cloud's performance management challenge
Leveraging cloud-based technologies, whether private or outsourced, for delivering
business-critical applications backed by solid service-level agreements (SLAs) allows IT to maximize the value of
the network by enhancing internal processes and operations. This is achieved via higher server utilization rates and lower costs thanks to
virtualization, automation and service management software.
The latter is particularly critical for enterprises deploying their own private cloud or leveraging cloud
services because the convergence of compute, network and storage streamlines the provisioning and allocation of
resources. For example, when the business specifies the application and class of service they want, IT should be
able to provision the appropriate resources automatically. The ability to plan capacity, monitor resource
consumption in real-time and make adjustments on-the-fly is also critical because IT must dynamically reallocate
resources depending on the level or class of service required by the application and how well the application is
Enterprise IT needs complete and immediate visibility if they are to meet SLAs with the business. With the
company's revenues and reputation at stake, they must ensure they are meeting business expectations while
optimizing their infrastructure to scale across a multitude of end users and applications.
IT organizations require highly automated and integrated system management software tools to optimize IT
processes and application performance across dynamic hybrid cloud environments
that combine on-premise private clouds and off-premise public clouds with traditional and virtualized IT service
infrastructures. Many are opting for a phased migration using technology partners and starting with less
mission-critical applications and resources, ensuring they have the people, processes and the right tools in
Keeping the lights on during transition
Traditional tools employed for managing performance have typically been point solutions supporting a specific
technology area. Because cloud services are delivered as one offering, the ability to collect data from different
sources, overlay that data in one interface, and view performance metrics as a whole becomes extremely important
for identifying and addressing service impacting issues.
But before organizations can identify what applications can or should be run in the
cloud, they need to understand how applications perform and the resources they require. By monitoring and
baselining performance, organizations can understand the peak utilization hours across resources in order to
determine the appropriate services or resource categories that must be defined within the cloud to support the
Once the apps have been migrated, IT needs to continue to monitor the cloud for changes or deviation of
baselines to determine if more resources are required. In some cases, resources can be dynamically applied at
certain times of the day to keep up with peak demands, then released when those peak times have passed, making
those resources available for other applications.
Leveraging virtualization in the cloud enables organizations to dynamically reallocate or move virtual machines
(VMs) to resources when needed. However, moving VMs to different resources can present problems for IT in terms of
maintaining visibility of the VM and assessing the impact of the move, because some tools will treat a moved VM as
new and history can be lost.
Solving the cloud management conundrum
Enterprise IT organizations leveraging private and off-premises public cloud services face several additional
performance management challenges:
" Cost containment -- Collecting, monitoring and baselining performance across technologies is critical to
allowing IT to right-size environments based on the needs of the services they support and deliver, therefore
optimizing or maximizing their infrastructure investments.
" Timely notification -- Baselines are essential because IT must rapidly identify changes in performance
behavior. Business dissatisfaction is a big problem and when performance issues arise due to an increase in
application usage, proactive notification allows IT to dynamically reallocate resources to ensure SLAs are met.
" Traffic management -- Analyzing usage data allows IT to better architect and plan for new services. They need
to understand where they can reduce costs through virtualization and convergence of infrastructure, while also
enabling leading edge delivery models supported by usage-based accounting.
" Triage -- When problems arise, IT needs the visibility to quickly identify where within the cloud performance
of key business services is affected.
cool tools for cloud management
For IT, the primary objective is managing the infrastructure and connectivity and ensuring that the bandwidth of
the latter is sized appropriately so that SLA commitments on performance and resource availability can be met. The
ability to on-board new business applications and end users rapidly is a necessity, meaning a single performance
management platform is vital if IT organizations are to avoid the cost and complexity of having to deploy and
configure multiple systems and then aggregate performance data back to a single point to achieve end-to-end
Meanwhile, end users will also expect the enterprise IT team to be able to deliver end-to-end visibility on
performance and quality of the application services if they are to justify their cloud investment. Although SLAs
can be hard to enforce within a unified fabric scenario, moving forward, it will be essential for enterprises to
have the right tools to be able to quickly isolate and resolve issues before they result in major business
Cloud computing and
virtualization indeed are transforming the way in which IT is delivered and consumed. But as adoption continues to
rise, organizations require a unified view of network, server and application performance across physical and
virtual infrastructures to meet current SLAs and future demands.
Cruz is director of product management and marketing at SevOne.
Read more about data center in Network World's
Data Center section.