Platform as a Service Versus the Newton

September 18, 2013, Storage Developer Coference, Santa Clara, CA—Gerald Carter, EMC-Isilon compared and contrasted platforms as a service and the Newton tablet from Apple. The challenges facing PaaS are similar to those facing the Newton.

The problem for Apple was that the Newton was too early for the technology and markets. The technology limits for processors and memory severely limited the capabilities, which led to massive customer rejection of the concept.

Although the present situation is similar for platform as a service, it will surpass infrastructure as a service for consumers some time in the future. The apps will make money while the operations will cost money. The platforms will focus on things and doing rather than the systems.

All of this change depends upon the concept that compute is a function of data and data transformations. Business success is partially due to the ability to make good and timely decisions based on that data. Being operational effective, the ability to do things well, is not a strategy for success. All of the hardware is converging into an indistinguishable fog.

The driver for the change to a platform-based structure is the cloud. The adoption of cloud technologies is following the classical segmentation of innovative early adopters, early majority, mainstream, late adopters, and laggards. The innovators are always going to be the first to try new technologies, and the early adopters are now starting to use heterogeneous computing technologies. The early majority is looking for safe and secure solutions and is considering various cloud technologies for deployment in the next couple of years. The late and trailing groups, comprising about a third of the population, is content with the slow and steady solutions.

PaaS will overtake IaaS due to talent and scale. The driver is the difference between core functions and context. Business differentiated apps are much more valuable than the infrastructure they run on. Although there is a growing need for storage engineers, the move to cloud solution technologies limits the focus on the hardware. A public cloud makes the IT functions someone else’s problem. Platform developers greatly outnumber storage companies and engineers. To these people, the infrastructure has to exist, but selling the underlying hardware to these people is very hard.

The first efforts at PaaS were offerings like Force.com and Google’s app engine. The app engine transformed to Google Apps. Microsoft introduced Azure and the .net framework, only to back out when they saw that the uptake was not there. Administrators want control and individual machines, so the IaaS world has taken off in incarnations like the latest Microsoft Azure, Amazon Web Services, Rackspace, etc.

Now the technology environment is facing new trends which are predictable, and the associated disruption, which is unpredictable. The good part of disruption is that you only need to know it is coming at you some time in the near future. Influencers rated from a low to high scale include momentum, existing technology, ecosystems, new technologies, and business models. One certainty is that the technology will always catch up to the leading edge. Technology prediction is always hard.

The increase in virtualization helps move hardware into apps and services. Software moves into storage and the storage changes from block to file- and object-based storage. All of the technologies move into (commodity) platforms once the technology is mature.

The next generation of PaaS will include services like RDMS, NoSQL, email, etc. running in a common runtime on virtual machines. The code for these functions will be portable. PaaS will probably have a platform war associated with the various operating systems, and mobile apps development is going to be a major challenge. One alternative is to use a Rest-ful interface and device-specific user interface code. The software storage solutions will operate through distributed file systems.

The industry will continue to consume, reduce, and react to changes in technology over a long time span that starts in a virtual yesterday and goes through to a tomorrow. In some respects, the iPhone demonstrates this principle as a re-visitation of the Newton with newer technology. Tomorrow is not a linear extrapolation of today, but may be a redeployment of yesterday.