Developers move towards the 'I’m-In-Charge’ application
- Summary:
- Sir Tim Berners-Lee pointed the way at the recent IP Expo, but the event showed that all the components of a new application- and data-centric way for the Cloud and IT are now in place, and coming together.
During his keynote presentation at the recent IP Expo conference in London, Sir Tim Berners-Lee talked at some length about the future importance of building data infrastructures. Most of the core mechanisms that are likely to be used to create this are now starting to slot into place, but seem to suggest a slightly different, richer perspective on what might actually appear than just new data infrastructures.
Put glibly, what may be coming in the direction of developers is the rise of the 'I’m-In-Charge- Of-My-Own-Destiny’ application.
During his presentation Sir TB-L, talked about Big Data, but stressed that the far more important aspect was 'rich’ data. Getting to this, and the value that can then be extracted from it, will involve a number of factors which are now starting to appear.
The key three factors are the move towards a world of Software-Defined Everything (SDE), coupled with the increasing power that comes from intelligent use of APIs, and the arrival of software and hardware systems technologies such as that from Nutanix.
Together, they create an environment in which data and applications can be dynamically merged into increasingly `virtual’ enriched datasets that can be processed just about anywhere, and on any system, that is appropriate for that work to be done.
Under way
The move to SDE is still under way, of course, with Software Defined Networking fairly well established now, and Software Defined Datacenters really starting to gather a head of steam.
As Clive Freeman, Chief Technologist with HP’s Enterprise Group told an IP Expo seminar, this started life as a way of containing the costs associated with managing the rapid growth of virtual servers, but has soon developed into a place where the agility of the compute component is the key to making it work well.
In an enriched data environment the compute component has to be able to change to whatever state the application requires – such as high compute performance, high data throughput, or cost conscious operations. At present making such changes are still an additional management process where a call for application X to run first requires a `third party’ management tool to spin up the required servers with a pre-optimised configuration.
Neil Fenton, chairman of 10Duke, pointed out at IP Expo that future applications will carry this information themselves, together with the intelligence needed to perform the necessary set up and have it runnable – such as auto scaling, on-boarding and off-boarding data as necessary - on any appropriate environment.
Companies such as 10Duke and MuleSoft are pushing the growth and versatility of APIs. This will play an important part here, as this is where the enrichment processes will normally occur for data. The ability to engineer business processes that are increasingly specific will follow as APIs allow applications to collaborate more often and more deeply, a trend which will occur as new applications are developed with API-based collaboration possibilities firmly in mind.
This, in turn, will create `virtual’ sets of dynamically combined data, where the individual datasets are not changed or corrupted, but an enriched dataset becomes available as part of the collaborative processes of the applications. This is likely to create new business opportunities built on the fact that the `whole’ will be worth more than the sum of the parts.
The final part of this is the new commodity hardware that will be coming along – the increasing trend towards very high volumes of small, commodity `application appliances’ rather than large servers running increasing numbers of virtual server instances. This is a market Nutanix is moving towards, according to company Senior Vice President of Marketing and Product Management, Howard Ting.
While its initial target markets are very specialised, such as providing the hyper-converged, synchronised clusters behind UK-based pi’s SaaS-delivered Enterprise Decision Support services, Ting see huge growth coming in webscale service delivery technologies. This is the world of appliances, millions of them, where each one runs an individual task that self-configures the appliance resources to its own needs and then removes itself and its data from the appliance.
These individuals can be converged into whatever larger resource – both of compute power and applications collaboration – a business process requires. And because it can be dynamically converged, the resources used can be the most appropriate – from rarer, more functionally specialised appliances that are still on-premise for the user, through to public cloud resources on the other side of the globe if costs, data sovereignty, security and process latency issues deem it appropriate.
Increasingly, these webscale appliances will also be designed to process data in-memory, as a commodity technology, allowing processes from the smallest to the largest to run in the same hardware…….except that it will be dynamically re-configurable and convergable from a single appliance to all appliances in the world, with combinations of collaborating applications pulling the strings.
My take
The core pieces of this model are now in place and coming together and are likely to have an increasing impact on the way IT is perceived and exploited, and is likely to move towards a higher level of abstraction away from the technologies themselves.
