Converged IT infrastructure and revolutionary homogeneity based on Intel’s industry-standard x86 architecture are described in this 2012 marketplace update from Pund-IT’s Charles King.

1. Marketplace Update
January 2012
 Revolutionary Homogeneity:
The Benefits and Evolution of
Converged IT Infrastructures
By Charles King, Pund-IT, Inc.
Pund-IT, Inc. Contact:
Hayward, CA Office: 510-383-6767
U.S.A. 94541 Mobile: 510-909-0750
charles@pund-it.com
www.pund-it.com

2. Revolutionary Homogeneity: The Benefits and Evolution
of Converged IT Infrastructures
By Charles King, Pund-IT, Inc.
For decades, heterogeneous computing systems have represented the rule of thumb in the
vast majority of enterprise data centers where a plethora of servers, storage arrays and net-
working switches hum amid a swirl of multi-colored cables. Here, growth can be remarka-
bly non-strategic. Systems are chosen according to application/workload requirements,
management or administrator preference, winning proof of concept (POC) bids or fire sale
pricing by preferred vendors and VARs. As a result, most facilities are far more complex
and far less efficient than they could be.
Would homogeneous systems provide better options for business computing? Actually,
yes, but such solutions typically consisted of proprietary systems whose vendors pitch the
value of working with a single product family or architecture—simplified administration,
more efficient data center management and “one throat to choke” if or when something
goes wrong. That isn’t necessarily inaccurate, but it qualifies as a remarkably myopic and
self-serving vision of data center nirvana.
However, continuing technological and market evolution have given rise to a far more revo-
lutionary and potentially more effective sort of homogeneity; what most people refer to as
“converged IT infrastructures” where all of a data center’s assets are based on Intel’s indus-
try standard x86 architecture.
The x86 Convergence
This may seem like an odd or even unlikely concept. After all, while x86 certainly domi-
nates corporate desktops and powers the vast majority of server units sold annually—
market prognosticators peg the number at 90% or more—the architecture is small potatoes
in storage and networking. Right? Well, only half-right. During the past half-decade, storage
vendors have increasingly abandoned proprietary silicon in favor of commodity CPUs for
array controllers. In fact, only one Tier 1 storage vendor has failed to make this transition.
Additionally, x86 is crucial to so-called cloud-optimized storage; the massively scalable,
highly automated systems supporting rapidly growing cloud computing infrastructures.
Networking, however, is a different situation today, with most vendors continuing to lever-
age custom ASICs in switches, routers and other fabric components. However, we expect
that a similar evolution towards x86 is likely underway. Plus, given continuing market/
pricing pressures and increasing commoditization (the same drivers that led to the success
of x86 in servers and storage), industry standard CPUs becoming the default choice of net-
working vendors seems highly likely. Finally, Intel’s increasing focus on developing innova-
tive networking solutions, including its July 2011 acquisition of Fulcrum Microsystems, a
leading player in semiconductors for high performance, low latency 10GbE and 40GbE fab-
rics, is likely to hasten this transition.
2

3. Why x86?
To understand the value of x86-based solutions in future networking and converged data-
center scenarios, it’s worth considering how the architecture came to its server market
leadership position in the first place. First and foremost has been the continual, upwardly
mobile evolution of x86 technologies. Originally launched as an embedded computing so-
lution in the late 1970s, x86 was ideally positioned to participate in and take full ad-
vantage of the various booms in desktop computing that followed. But as IT became in-
creasingly decentralized, x86 also substantially enabled and profited from those develop-
ments.
By the early 1990s, x86-based systems were firmly established as cost-effective options for
edge-of-network Web servers, file/print and similar applications once dominated by propri-
etary RISC servers. At the same time, developments in clustering, grid and other areas led
to x86 being leveraged in high performance and supercomputing scenarios, another mar-
ket now dominated by commodity component solutions. Then in 1999, VMware introduced
virtualization for x86, dramatically improving system utilization and providing organiza-
tions the means to consolidate multiple workloads and applications, and reduce the num-
ber of servers they managed.
These steps all contributed to x86 becoming the default architecture for cloud computing
pioneers, including Google, Yahoo, Amazon and Facebook, and the vast majority of today’s
commercial cloud solutions leverage industry standard CPUs. At the same time, x86 of-
fered traditional system vendors the means to develop ever more powerful, capable solu-
tions while preserving or improving their profit margins. But just as importantly, x86 has
been critical in the emergence of non-traditional systems vendors like Cisco and VCE, and
the development of innovative, dedicated appliances for Big Data, data warehousing and
analytics, virtualization and data center consolidation.
To the Network and Beyond
There’s been a similar evolution since the arrival of x86 on the storage scene. Along with
allowing storage vendors to enhance their solutions’ performance and profit margins, com-
modity CPUs have provided the means for developing entirely new integrated systems that
closely couple highly virtualized x86 servers with a few petabytes of disk. Not only does
this approach offer cloud service providers the means to easily build-out cloud infrastruc-
tures and scale resources as needs require, but it also allows storage vendors to focus their
time and resources on developing value-added software and services.
We expect a similar transformation to occur in networking and are already seeing signs of
it catching on as companies recognize that cloud data centers which essentially become
highly integrated, automated and optimized “computers” need new ways to communicate
between server nodes. Google was one of the first companies to pursue this vision but it is
also apparent in Facebook’s Open Compute project. Both use virtualization to effectively
separate the software that controls where packets go from the switches that perform physi-
cal routing.
3

4. Intel thoroughly recognizes the value of this approach both to itself and its various OEM
customers. The company’s acquisition of Fulcrum Microsystems wasn’t simply a means of
ramping up production of new commercial solutions. Instead, it recognizes the networking
community’s steady movement toward merchant silicon and qualifies as the next logical
step in the journey toward converged data centers which comprehensively leverage Intel’s
Xeon architecture across all components.
Final Analysis
The phenomenal success of Intel’s x86 architecture in servers, its rapid adoption in storage
arrays and its bright future in networking fabrics should eventually lead to fully converged
data center solutions, a development which will benefit both established and emerging sys-
tems vendors and data center owners of every stripe. In essence, x86 provides vendors and
service providers the means to compete effectively, while differentiating their solutions
with value-added software and services.
Overall, we consider these developments emblematic of Intel’s long history of commodity
silicon innovation and reflective of the company’s deep understanding of IT industry
trends. By enabling and encouraging truly homogeneous IT components, Intel is also help-
ing to make the dream of cloud computing a common reality for vendors and their custom-
ers.
© 2012 Pund-IT, Inc. All rights reserved.
About Pund-IT, Inc.
Pund-IT emphasizes understanding technology and product evolution and interpreting the effects these
changes will have on business customers and the greater IT marketplace.
4