Why is the cloud still so foggy?
14 January 2016
In my opinion, ‘in the cloud’ means ‘a computer server’ in ‘some’ data center ‘somewhere’ on the planet offering ‘some kind’ of service that ‘someone’ or ‘something’ can connect to. This definition is a bit vague, but ironically also accurate. Nowadays, the trust placed in the ‘cloud’ working reliably is already pretty high and still growing. Let’s put it a bit into perspective and go back in time…
The very first wide spread use of computation based office work were users logging into a mainframe server, somewhere in a big room in the building, through simple serial connected terminals. Most of them spent all their computer time behind that terminal and never asked themselves how it worked. For the terminals no expensive IT guys, license management, updates etcetera were needed because the terminal only had minimal requirements. In essence they were working on the mainframe, but their terminal was just the bare minimum to serve as a user interface (screen, keyboard and ‘maybe’ a mouse). They had no clue where data was stored, how it was stored and backed up and maintained. Actually most had never ever seen a mainframe in their careers. People saw the benefit and also wanted one at home. Unfortunately a mainframe back then was huge, expensive and needed specialists. Furthermore it was not economically viable to have serial communication between the home and the office to make a terminal work for home use.
After quite some years, the home computer was invented for that purpose and most of us know the stories about the rise of Microsoft, IBM and Apple computers. You have your ‘own little mainframe’, just for yourself, which gets more and more powerful over time (Mhz to Ghz and Kbyte to GigaByte). The euphoria of doing things at home slowly changed to more and more of a headache, because computers were getting too complicated to maintain, needing expensive licenses for operating systems and applications, backup issues, new motherboards and faster CPU’s to cope with the workload.
Let’s talk about the cloud now.
Consider Salesforce, around 2001 one of the first companies to offer online enterprise class applications. Numerous companies use them and trust them with their most precious information and for vital daily operations. The ‘failure’ of Salesforce would most certainly mean the ‘failure’ of 150.000 other companies (based on the 2015 annual report). Another example is Amazon, offering internet storage space and computation facilities for 3rd parties since 2002. These are just 2 well-known companies doing ‘cloud’ based stuff for more than a decade. When I used Hotmail for the first time, my mail was in the ‘cloud’. When I did my first payment with my ATM card, my bank account was in the ‘cloud’. When I uploaded my birthday picture on FaceBook it was in the ‘cloud’.
Then why is it such a hot topic now?
Aren’t we doing the same ‘mainframe’ thing now and calling it ‘in the cloud’? In essence, it looks like things seem to be rephrased as if it is something new.
But wait a minute, is that it? Basically yes, however the proliferation of reliable and fast internet everywhere makes it a lot easier to use the cloud (again meaning: datacenters) than ever before. Powerhouses like Google, Microsoft and Amazon are now exploiting their huge presence and know-how in maintaining high-end globally connected data centers. They take care of the infrastructure, servers, uptime, backup, security, scalability and they are prepared for huge volumes of data and connections.
Now how about using the cloud in the industrial arena?
Since the industrial automation world (and especially the process industry) is quite conservative, right now you predominantly see the use of the cloud to collect and display sensor and telemetry data and not yet for control data (data which directly manipulates the process). What is interesting is that diagnostic information of network technology (e.g. PROFIBUS and PROFINET maintenance information) is a very good next candidate to be one of the first focus points of putting industrial information to the cloud. As it is ‘read-only’ information, arguments like security are not really an issue. Think about it: Who really cares about somebody else seeing how well your network works?