
Virtualization and Cloud executives share their predictions for 2017. Read them in this 9th annual VMblog.com series exclusive.
Contributed by Ashok Rajagopalan, Head of Product Management, Datera
Four Things to Consider in 2017 - The Year of Cloud & Automation
Over the past few years, the benefits of cloud computing, be
it on public or private cloud, have become clear, yet many enterprises still
struggle to operate workloads and manage their data at scale. The key to
enterprises achieving simplified management at scale is automation and
decoupling applications from physical infrastructure.
A recent enterprise cloud are study by VMware discovered that a typical
enterprise IT team operates at least eight individual clouds. Another peril of many
of these clouds is that each one has a unique operational model, so
orchestrating and managing all workloads and associated data warrants a massive
set of IT resources and time, or a new operational paradigm driven by
automation and abstraction. . While, a number of new workloads orchestrators
such as Mesos and Kubernetes have started to deliver the abstraction for
compute and networking, there aren't many such solutions delivering similar
capabilities for storage. Hence, containers and their ecosystem adoption has
grown primarily for stateless workloads, and are lagging for stateful
workloads.
Modern clouds have become synonymous with intelligent
software running on commodity x86 hardware. It is even more imperative that
this software handles all the organic lifecycle process for the cloud,
including software and hardware upgrades, scaling up and down, reconfiguration
and patching. A truly well-defined cloud infrastructure should shield the
application from impact from these processes.
Gartner predicts that cloud infrastructure will
impact one trillion in IT spending making the tenants of cloud infrastructure and automation around its
operations critical for success in the enterprise. While automation primarily deals with
consistent processes leveraging the abstraction of workloads from cloud
infrastructure, there are some core principles required to make it effective.
Here are four things enterprises should think about as they
consider implementation of an automated cloud infrastructure in particular for
the data infrastructure.
1. Policy: With an explosion of data
growth, enterprises need agility and application-level control to provision and
manage data through its lifecycle. Automated provisioning with
application-level controls such as access and security, performance (IOPS,
bandwidth and latency) and data services, coupled with scale make it
foundational for automated data infrastructure. These inputs need to self-drive
the data infrastructure if the IT operator has to be relieved from mundane hand
crafted provisioning and management. Clearly-defined, sophisticated policies
should be the first thing companies look for when considering a new solution.
2. Performance: One of the most common
criticisms of shared infrastructure is
that performance in particular latency is higher than on dedicated alternative.
On-premise cloud infrastructures specifically can address these concerns.
Taking advantage of low-latency data fabrics, and modern media types (such as
NVMe flash, NVDIMMs), application latency can be significantly reduced, more so
than dedicated infrastructure. In concert, workload isolation techniques such
as QoS, multi-tenancy, and micro-segmentation gives control to the application
owner. According to a Taneja research survey of IT buyers and
practitioners, 40% of respondents identified the ability to better control and
meet application service level commitments (SLAs) as a top priority.
3. Security: Typically, security is
associated with encryption. With the cloud, access controls and segmentation
are as important to achieve security. It is imperative that these modes of
security are built into the operational aspects especially with the persistent
data layer in the cloud. This ensures that as workloads scale, with automated
provisioning, security enforcements are intrinsic and real-time to the solution.
4. Integrations: For cloud automation to
span beyond the orchestration layer, integrations with other elements such as
persistent storage layer, monitoring and event handling services, image
management services, and so on, are vital. For example, in Kubernetes, the data
infrastructure layer can be integrated using the FlexVolume framework to make
provisioning seamless and automate to scale.
IT roles will shift dramatically over the next year as the
tasks that they've traditionally managed took so much of their valuable time
will be automated thus allowing IT professionals to manage more strategic projects. By
thoughtfully selecting and implementing cloud solutions, enterprise
applications can smooth that transition and enjoy the benefits sooner.
##
About the Author
Ashok is responsible for product direction and strategy at Datera. Prior to joining Datera, Ashok was a key part of the product team that brought Cisco UCS to market, and ran the UCS applications and cloud product teams that conceptualized FlexPod and Cisco Analytics platforms. Ashok started his career as an I/O and OS architect in HP-UX at HP. Ashok holds an MBA from UC Berkeley and MSEE from University of Arizona, Tuscon, and a number of patents in the IO and resource mgmt space.