Infrastructure Performance Management: 6 myths debunked

Application Performance Management (APM), Network Performance Management (NPM), and Storage Performance Management (SRM) tools give a view of the Applications, Network and Storage respectively but what companies want is a single view of how the infrastructure is performing as a whole. Much time and effort is wasted trying to optimize a system by looking at each element in turn – the effort increases if there is a slow-down in an application and a resolution is needed in a hurry.

Many suppliers have a vested interest in not improving the utilization, and therefore cost, of the infrastructure and so myths have emerged to ensure you stay with a less efficient system. This article debunks some of these myths and gives you some infrastructure facts and questions you should be asking:
£ Myth: Overprovisioning is the only way to effectively manage for
performance spikes.
£ Reality: The days of overprovisioning to accommodate bursts
or changes in application requirements are long gone. With
data volumes growing at 30-50% a year, CIO’s are driving their
infrastructure teams to fully sweat the assets on the floor. That
means IT must ensure they are running at just enough capability
to meet the requirements of the business.
£ Myth: Utilization equals performance.
£ Reality: Utilization measures the percentage of capacity in use
by either servers or storage. However, an infrastructure can have
excess capacity – i.e. be underutilized – but still not be performing
well. Utilization was an effective measure of what might impact
performance when organizations had the luxury of overprovisioning
resources. Utilization is no longer sufficient to understand how
the workload changes affect the overall performance and end
user experience.
£ Myth: Managing individual devices or components adds up to
a view of the end-to-end infrastructure.
£ Reality: Large enterprises run complex, heterogeneous
infrastructures that rely on multiple vendors. Device-specific tools
collect, measure and view data only from their point of view. They
don’t tell the whole story. It’s impossible to take metrics from each
device and add them up to get an accurate view of the real-time
performance of the systems across the hypervisor, server, system,
network and storage layers. The individual metrics are still
important and can provide insight into the overall performance
and availability of the infrastructure. However, they need to be
aggregated and correlated by a platform that provides an unbiased
view across the end-to-end system.
£ Myth: Polling in 5 to 20 minute increments is adequate in today’s
virtualized and cloud environments.
£ Reality: Today’s infrastructures, many of which rely on virtualization
and cloud computing, are more flexible and dynamic than ever
before. So much is happening within a couple of minutes that
polling technologies, which gather data in 5 to 20 minute intervals,
are no longer sufficient. By relying on polling technologies alone, it
is easy to miss transient problems – or ‘ghosts in the machine’ –
that can have a big impact on the business.
Organizations need to measure performance in real time – at line rate. That means measuring to the microsecond and rolling up and delivering data in one-second intervals. Data needs to be continuously captured at extremely granular or finite precision to get an accurate view of the overall infrastructure performance.
£ Myth: Tapping the infrastructure negatively impact performance.
£ Reality: A Traffic Access Point (TAP) is a passive, out-of-band
optical splitter that divides the light traveling down a fibre optic
cable. TAPs are quickly becoming a best practice in today’s highly-
virtualized data centers and we are seeing industry leaders, such
as Corning, enter the TAP market. Tapping the storage area
network (SAN) enables IT to monitor the physical layer and analyze
the fibre channel at the protocol layer. The real-time data that is
captured helps ensure higher performance, improved reliability and
better utilization of the overall infrastructure.
Many of the world’s leading companies, that can count the cost of application latency or downtime in hundreds of thousands of dollars per minute, TAP all their SAN ports as the cost of implementing TAPS is far outweighed by the advantage and cost savings of fast reaction to issues and the ability to optimize the entire infrastructure.
£ Myth: Application Performance Management tools measure
infrastructure performance.
£ Reality: Application Performance Management (APM) tools
are complementary to Infrastructure Performance Management
(IPM) tools. APM solutions monitor and manage the performance
of applications while IPM solutions looks at the performance of
the infrastructure as a whole – from the physical infrastructure to
virtualized and cloud computing environments. IPM solutions
provide a holistic view across the whole IT system and give context
to the behavior of individual applications. With more and more
interactions happening between the infrastructure and applications
it is important to have a 360 degree view of the infrastructure plus
an understanding of application performance.
Today’s Requirements: A Purpose Built Platform for Infrastructure Performance Management
A purpose built Infrastructure Performance Management (IPM) platform provides visibility into the performance and health of the physical, virtual and cloud computing infrastructures. An IPM platform needs to provide a view across the systemwide infrastructure – from the virtual machine to the logical unit of storage.
The key tenets of an IPM platform include:
£ Continuous real-time measurement: Data must be aggregated
and presented in real-time. It must be captured in a database and
be referenceable in a continuous manner.
£ Unbiased: An IPM platform must take a vendor independent view of
the whole system.
£ Heterogeneous: The ability to collect and correlate data from
a variety of devices in the infrastructure.
£ Systemwide visibility: The ability to track any activity across the
IT infrastructure.

With an IPM platform in place, IT can right size the capacity of the infrastructure, increase resource utilization, improve application response times and proactively identify and remediate problems.
This allows IT to establish business aligned service level
agreements (SLAs) and to be aligned with the overall business requirements.