The Gathering Cloud(s) of Creative Destruction

Written by Chuck Ward

aNewDomain.net — IT has seen huge growth in server virtualization, allowing the consolidation of multiple physical servers into a single physical server running multiple “virtual” servers. And not surprisingly, this consolidation has had a huge impact on enterprise IT.

As an example: American International Group (AIG) recently published a report about the consolidation of its data centers running on an SAP platform. It went from 28 data centers down to two (a reduction of 92 percent), consolidating 12,000 – 13,000 servers into 2,500 (a reduction of 80 percent), and saved a projected $1 billion. Of course this transition (virtualization) requires powerful servers and specific virtualization software, both of which are hugely expensive. However, if the reduction in staff and energy costs (which were not mentioned) are significant, AIG could follow the same 80-90 percent reduction. This is a textbook (or Wikipedia) definition of “creative destruction,” and this is just the beginning. The move to the cloud promises to be an even greater phase of creative destruction.

640px-Dark_storm_clouds

Image credit: Wikimedia Commons

So what will the enterprise industry look like as its customers move away from their own, local enterprise IT infrastructure toward the cloud?

What happens to the Infrastructure?

The Open Compute Project, formed in 2011 by Facebook to openly share its designs to create the lowest cost, has a plan to present the most efficient data center. Its ambitious goal is well defined in its Mission and Principles statements and seeks to open source the entire data center, including physical layout, cooling, power, hardware, and even the software stack.

Servers are built from commodity components, and cost is reduced by omitting unnecessary components such as expansion slots and IO connectors. Costs are also kept down by using metal rack mount trays that forgo fancy cases. In this way all the components are exposed to natural air cooling, and allows for quick and easy replacement of faulty components.

When a company as big as Facebook decides to design and build its own infrastructure (it claims to have close to 200,000 servers), and openly share and invite others to join in, people tend to listen. More than 50 companies (including Intel, ASUS, Dell, Huawei, Red Hat, Cloudera, Mozilla, and Rackspace) joined the group. This may be one of the most important decisions these companies have made in an attempt to remain relevant through this massive transition and retain customers.

What happens to the Channel, VAR, and Service Providers?

A lot of what the preexisting channel provides is hardware and advice on which hardware to purchase. But as more and more of this information moves to the cloud, it will significantly impact companies such as as CDWInsightPC ConnectionPC MallZonesTech Data, and Ingram Micro. These businesses will undoubtedly encounter new-found challenges when the products they distribute start showing up in the cloud.

What happens to desktops and packaged software?

As applications move to the cloud, the necessity for high-power/power-hungry PCs and laptops will decline. Consumers are moving toward mobile devices that access applications through a web browser. And they are quickly abandoning the bulky systems they previously used.

Recently Adobe took the bold step to move its most-popular application to the cloud. The company announced that it will no longer be shipping shrink-wrapped versions of its flagship Creative Suite (priced anywhere from $500 to $2,000). Instead, it is now offering a monthly subscription version of Creative Cloud at $50.

Microsoft has taken a more conservative approach by offering both a traditional shrink-wrapped Office 2013 ($140 – $400), as well as  its cloud-based Office 365 for $10 per month.

Adobe already offers its apps for use on mobile devices, but Microsoft will not release Office 365 until 2014.

Per use or subscription cloud-based applications such as Google Docs will reduce or eliminate demand for shrink-wrapped software and expensive licensing deals and pose a significant challenge for application vendors in the future.

What happens to “Intel Inside”?

ARM Holdings‘ low-power, low-cost RISC technology design is perfect for portable electronics (like smartphones, tablets and other products).  Rather than manufacturing its own chips, the company licenses its designs to other companies to produce. In addition to mobile devices, ARM is seeking to bring its products to both the server and the desktop markets. Even Microsoft has developed a version of Windows 8 to run on the ARM chip.

As more users adopt ARM-based mobile devices instead of a PC, Intel and AMD have obviously encountered this challenge and are working on new chips to compete with ARM, generating speculation that Apple may consider using Intel chips in future iPads.

What happens to storage?

Higher CPU utilization means higher storage utilization requirements, forcing shifts in what type of storage is used, and how it’s used. Tape, hard disk drives (HDD), and even standard solid state drives (SSD) are far too slow to keep up with the demands of this new architecture, measured in IOPS (Input/Output Operations Per Second).

Companies such as Fusion-ioOCZ Technology offer what is known as “persistent memory,” which blurs the line between storage and memory and enables far greater performance than SSD or HDD storage. It also pushes traditional SSD and HDD to the bottom of the storage stack where tape had been before. This shift will have a highly-disruptive effect on the world’s two remaining hard drive vendors (Seagate and Western Digital). It will also adversely effect storage vendors, like EMC and Network Appliance because their products will offer greater performance (IOPS) at a lower cost.

What happens to Energy consumption?

In a 2012 report, General Electric cited that data centers consume two percent of all electricity generated (38 billion watts), costing $2 billion per month, with CO2 emissions greater than the total CO2 emissions of countries such as Argentina, Netherlands, and Malaysia. One metric used to benchmark the efficiency of a data center is the Power Usage Efficiency (PUE), an index of how effective the entire data center is at using electricity for its operations (including power for servers and cooling).

Virtualization increases the efficiency of the enterprise data center by enabling more than one application to run on a physical server, therefore reducing the number of physical servers.

In an effort to maximize efficiency and reduce their carbon footprint,  Google, Amazon, Facebook, RackSpace, Apple and Microsoft are all locating their data centers in areas of the country close to sources of renewable energy and natural cooling. These data centers are even built to maximize cooling from wind patterns, as well as optimal positioning for wind turbine and solar energy generation.

Conclusion

In his 1996 book Only The Paranoid Survive, former Intel CEO Andy Grove wrote about what he termed a “10X change,” a point at which a company (or industry) comes face to face with a massive change, one powerful enough to threaten the life of the company. In a time when virtualization and server consolidation have significantly shrunk the enterprise data center, the greatest disruption will come when enterprise customers move entirely to the cloud.

For the past 10 years Chuck Ward has been the CIO for a private investment firm, and he is a valued contributor to aNewDomain.net. In addition to his day job, he is also the founder of both The Hartford Innovation Center and Stealth Mode Projects–incubating new ideas and mentoring senior engineering students on their capstone projects.