Microservers, also known as Density Optimized Servers, are an important technology advancement that will play a significant role in server infrastructure decisions, but will not adversely affect the data center industry.
Microservers are a low powered trimmed down version of standard servers, requiring 10-40 watts as opposed to standard servers requiring 90+ watts. Their primary appeal is low power consumption and a small form factor, thus saving power and space.
The primary purpose of microservers is to accomplish simple, single task activities, like serving static html pages, which required a very low amount of processing or computer power. This limits the applications microservers can be used for and will therefore temper any widespread replacements of traditional servers, which offer a robustness not designed into microservers.
Further to these fundamentals, historic technology adoption trends support the conclusion that this new technology is not likely a threat to the growth of data center requirements in the future.
For example, blade servers were introduced in 2001, nearly 12 years ago. The primary appeal of blade servers is very similar regarding small form factor and lower power consumption. To date, blade server technology is still popular as a way of managing rack mounted server infrastructure, but after 12 years the blade server market represents only 16.3% of the total server market revenue. In that timeframe the data center market has continued to grow substantially.
Another prominent example is the server virtualization market. Server virtualization, although technically a technology that was developed in the 1960s, was not embraced by the overall information technology industry until 2001 when VMware introduced their first server virtualization product.
Virtualization promised to allow IT departments to consolidate multiple servers with low utilization into one server. The technology is so popular that every primary and many secondary IT vendors now compete in the space. There are limits to virtualization including scalability and inappropriateness for many critical workloads. To date, 40-50% of servers in mid to large enterprises are virtualized.
Even with the introduction and adoption of these and various other technologies, all intended to lower the cost of computing and lower data center requirements, we see the market for physical servers continue to grow, as depicted in the following charts.
2009 shipments dropped dramatically, as would be expected based on the economic environment. The dip of 200,000 units in 2012 represents a 1.5% drop in shipments, but the fourth quarter showed positive revenue growth for server vendors year over year. While these statistics are important to watch, there is no indication yet that technology advanced are responsible for the decline, but rather a cyclic event in purchasing patterns.
The following chart depicts number of microserver shipments as estimated by IHS. Even at the 2016 level of 1.2 million servers, this only represents about 14% market share based on 2012 shipments and will likely be lower by the time 2016 rolls around.
Conclusion: While the numbers and markets should be monitored for significant events, there is no evidence that lower power or smaller form factor technology will erode growth of the data center industry. In fact, the third party colocation data center market is largely driven by the growth of those companies that have already made the move to IaaS (infrastructure as a service), and those organizations already interested in consolidation and the decommissioning of the aging and less efficient data centers.