I was recently talking with a colleague about new technologies and noticed a pattern: Many of last year’s innovations in hardware are very similar to next year’s new software features. The more I thought about it, the more I realized that this IT trend is nothing new.
More on IT trends
Ten years ago, for example, server sprawl was a very real concern. As end users demanded more servers for a particular application or task, hardware vendors turned to small-appliance form factors and blade configurations. These new server technologies helped promote denser server environments.
At first, the need for a large number of small servers was met in the physical world. Not far behind this hardware innovation came server virtualization, a software answer to the need for higher server counts. While this need was initially easier to meet with hardware, software vendors soon arose with a more efficient solution.
Fast-forward to today and there are similar IT trends. Many networking devices -- from load balancers to firewalls -- are now available as virtual appliances. What once required purpose-built hardware can now be run on virtual hardware and deployed almost anywhere. In fact, virtual appliances are actually becoming the preferred format for networking equipment in many data centers.
When you think about, this leap frogging of technology is really just following Moore’s Law, which states that computing power doubles every two years. Initially, when a difficult problem emerges, purpose-built hardware is coupled with custom application code to create a device that fills a very specific need. Over time, computing power and scaling matures to a point where more flexible and dynamic software can replace the hardware solution.
IT trends in server storage
Server storage was the technology that initially got me to think about this pattern. Just a few years ago, I marveled at the hardware innovation in storage caching, with technology such as EMC Inc.’s FAST Cache and NetApp Inc.’s Performance Acceleration Modules. While these physical-storage features are still important, vendors are providing similar functionality at the software layer with features like VMware’s Content-Based Read Cache (CBRC).
In both the hardware or software realm, caching relieves the amount of data that needs to be read from the storage media. With hardware solutions, data still traverses an Ethernet or Fibre Channel network. With software features like CBRC, however, data is cached at the hypervisor level and doesn’t have to traverse an Ethernet or Fibre Channel connection.
There are even more storage examples: Pivot 3 created grid-storage solutions from local hard drives on which the hypervisor resides and Atlantis Computing's ILIO creates a NAS volume out of a local or shared disk and adds both dedupe and caching to that data store.
The takeaways from these IT trends
These cycles of innovation have shown that vendors will usually release purpose-built hardware first, when the solutions are in their infancy and resource consumption is more demanding. Over time, the solution will mature to lower resource demand, while Moore's Law dictates that the computing power available will improve until reaching a point where software implementations become feasible. Even beyond this point of conversion and the introduction of software options, hardware-based tools continue to drive innovation and remain the solution of choice for higher-performance implementations.
While less nimble and more expensive, hardware vendors will likely lead innovation. Most of the compute resources reside with hardware, which can be dedicated to new features as they mature and increase in efficiency. However, software implementations are more flexible and usually less expensive. Where hardware innovation will bring new features to the market, software will make it available to the masses.