Virtualization and Cloud executives share their predictions for 2015. Read them in this VMblog.com series exclusive.
Contributed article by Piyush Mehta, CEO of Data Dynamics
Transformation to Software-Defined Storage: A Word of Caution
There is a large transformation taking place in the storage
marketplace. It is a fundamental shift, with storage companies moving from
being proprietary providers of the physical disk to being the software-based
management layer that will manage heterogeneous storage devices. This shift is
changing the way companies position themselves and is disrupting the
marketplace, with new entrants giving a true challenge to the legacy "goliaths"
that dominate the marketplace.
The shift of focus to the software layer is a result of the
vast amount of data that has been and continues to be generated across all
aspects of storage. Specifically, unstructured and big data is seeing storage
growth more than double every two years, with the file- and object-based
storage market reaching $38 billion by 2017, according to IDC. The growth is
seen across all verticals, from financials and pharma to retail and manufacturing.
The reliance on data for applications to make intelligence-based decisions
makes storage the most essential component of the infrastructure stack.
This level of data growth, combined with its importance, has
led customers to look at means of reducing the aggregate cost of storage while
ensuring there is no compromise on the performance demands of the applications
that call on that storage. The balance of cost and performance has led to
enterprises looking for innovative means of managing their storage spend.
An area of focus that has led to the transition we are
witnessing is that the underlying disks are becoming more and more robust and
stable. With the evolution and price reduction in solid state disks and more
scale and stability in traditional SATA drives, performance can be garnered
regardless of storage vendor for a large majority of the estate. This
realization has led enterprises to a belief that "a disk is a disk is a disk," regardless
of the front-end storage vendor, and as such there can be tremendous cost
savings gained through moving from proprietary storage vendors providing the
underlying disks/storage to a generic disk/storage solution.
In order to leverage generic storage, enterprises require a
storage management plane that is able to meet all the operational requirements
for storage management while being able to support and make optimal use of the
heterogeneous storage beneath it. In parallel
to a shift from traditional proprietary storage silos to generic ones, most
enterprises have embarked on a hybrid cloud strategy, leveraging external
providers for burst, archival, or primary storage. The reduction in management
personnel, floor space, power and cooling savings in the
data center leads to cost savings that justify the shift to a hybrid on- and
off-premise storage model. Thus, the
software management layer must be able to talk to the cloud providers and be
able to integrate the management functionality of those providers.
Although this shift to a pure software world sounds like a
great idea, and theoretically is one, it is easier said than done. There are
years of operational processes put in place based on legacy storage
infrastructure. This web of touchpoints across the enterprise, from links being
embedded in a spreadsheet to applications that are hard-coded with specific
hardware information, makes an immediate "rip and replace" impossible, as that
would be disruptive and potentially detrimental to the business. Therefore, the
transition has to be one that allows for utilizing the existing operational
processes while overlaying those processes with a centralized management
capability.
From the vendor standpoint, creating such intricate software
is not simple. On the front end, it
needs to be simple and easy to use, providing for a single pane for everything
from provisioning and allocation to ease of data mobility, replication, disaster
recovery, backup, and archival management. On the back end, it needs to have
API integration with all the various vendors while ensuring supportability
across the traditional silos of storage, including block, file, object, Hadoop,
and tape-based storage. The API integration is not trivial, as most vendors
don't have robust API stacks. Although vendors are starting to mature their API
and provide for greater functionality through API versus through the command
line, there is still a long ways to go before this is robust and proven.
The other challenge is the ongoing interoperability against
new releases by hardware vendors and assurance that the API that functioned
across prior versions continues to do so with the new versions. Add to this the
complexity of integrating with cloud providers to support a hybrid cloud
environment. When you take into the account all the various layers to account
for, the integration requires meticulous focus and ongoing maintenance.
Another key aspect to be cognizant of is "vendor lock-in".
For years we have seen storage hardware vendors tout the value of
virtualization and the ability to be non-disruptive. This virtualization layer
never gained the market adoption that was envisioned at inception because it
became a "lock-in", whereby the customer is unable to switch from the incumbent
vendor to another due to the dependencies created and the lack of visibility to
them. This same "lock-in" is a challenge with storage management software. The
software must provide transparency as to where exactly the storage is residing
and how it is deployed.
This information is vital to ensure that, should you choose
to remove the management layer for any reason, you are not blocked from doing
so due to the "black hole" of interdependencies on the back end of which you are
not aware. Enterprises have experienced tremendous pain when virtualization
needs to be removed and have seen the disruption lead to either running the
virtualized layer in an end-of-life or unsupported state or going through a
massive disruption that in most instances has taken several years to unwind.
There is no doubt that the transition from a traditional
silo-based storage infrastructure to one that is centrally managed through
intelligent software and can scale to meet today's massive data growth is
taking place. The process is one that over the coming years will drive
optimization within enterprise infrastructures and provide for an economical
means of managing the massive amounts of data, but the transition will be one
that requires caution and a tremendous level of planning and impact analysis
and be done in phases that ensure the least amount of disruption.
If
done correctly, the increase in storage capacity and its corresponding costs
will be countered by the efficiencies generated and the business advantages
provided by having access to a robust amount of data.
##
About the Author
Piyush Mehta brings over 19
years of entrepreneurial experience, with the last 16 focused on software and
technology services, to his role as CEO of Data Dynamics.
Having founded one company and nurtured another from commercial inception,
he had grown both entities to strong growth and poised them for further
success. As CEO at SANpulse Technologies, Piyush established alliances and
partnerships with key industry leaders within the storage industry, built a
strong management team, and developed a culture that inspired and rewarded
creativity and unity within the organization. Prior to SANpulse, Piyush founded
an IT services firm, Integration International, which he built into a
multinational entity and sold in 2006. He also held a senior-level executive
position with a multinational manufacturing company. He has extensive global
experience in treasury, investor relations, business strategy, acquisitions and
divestitures, finance, and operations.