The world is going software defined, of that there can be little doubt.
With servers becoming divorced from their operating systems (OS) through virtualisation, the same is happening to network switchgear and storage too has had the pooling development, so why is one of the key developments in enterprise IT going in completely the opposite direction?
Big Data has been a buzz term for quite some time now and the general noise has been that by being able to deal effectively with the volume and velocity of data coming into the organisation, advances in processing and analytics will allow organisations to gain new insights.
While many will be aware of this kind of thing being done in the likes of retail with loyalty cards, business analytics and Big Data analysis are being applied to all manner of data that may previously have not been looked at beyond its most immediate use, in all manner of verticals and business sizes.
However, one curious thing that seems to characterise the cutting edge of this trend is the tight marrying of hardware and software, the integrated stack as it were, that entirely bucks the trend of software defined.
All the arguments that have been put forward for virtualisation and the software defined approach should really apply here, but some of the leading companies in this area, such as IBM, Oracle and others, seem to be going down the engineered systems approach where there is a tight marrying of hardware with.
Now, that is not to say that there are no offerings for analytics in the cloud, but rather that at a certain point in the market, pretty much the upper end, the main options seem to be engineered systems that have the software firmly tied to the hardware.
Why might this be? Well it could be that the level of introspection required for this data requires a very specific level of performance that is difficult without having the direct control that these engineered systems offer, but as those now familiar arguments from the abstraction side of the house would counter, this means that there is poor scalability in this approach. Scalability is a key element in dealing with the velocity part of the three Vs of Big Data and so if an organisation embarks on this route, then lining up these expensive engineered systems would seem to be the only way of tackling an increasing amount of data to analyse.
On mentioning this conundrum to someone who knows about these things, a bigger picture emerged. Paul Daugherty, CTO, Accenture, reckoned that two models would emerge in this area, one being the engineered system and the other being cloud offerings. The two models that will have some tension between them going forward, said Daugherty.
The commodity oriented cloud model where it’s all about commodity hardware with very detached software, argued Daugherty, where you can abstract the software, spool up and spool down resources-that’s one architecture that’s very effective at solving certain problems.
The other side is the likes of IBM’s Pure technology, Oracle’s Exa-technologies and SAP’s HANA, there is a different orientation, said Daugherty, which is very specialised packaging and configuration of hardware and software together that in some cases can lead to tremendous efficiencies and improvements for certain problems.
Daugherty said that both are very viable models but there could be tension between the two models for a while. And users can be a bit confused by that as they look at moving things to the cloud and then on the other hand they hear about these engineered systems. It’s going to take a few years to play out, he said.
Both models are still a bit early in their development, Daugherty concluded.
So, despite the world going for abstraction, the anomaly of tightly integrated software and hardware in space will persist for some time as the two models mature and find their own users or expand their capabilities.
It may be nothing more than the right solution for the problem at the time, but it still seems a little odd that the Big Data phenomenon seems best tackled by the opposite of prevailing trends.
Subscribers 0
Fans 0
Followers 0
Followers