What is cloud computing? Is it, as the cynics put it, merely storing your data on someone else’s computer? Or is there more to it than that?
These are some of the questions self-described cloud technology inquisitor Daniel Tremayne-Pitter seeks to answer in the new documentary Clouded from Dark Matter, in association with Hewlett Packard Enterprise (HPE) and VMware.
Clouded charts Tremayne-Pitter’s “journey of cloud anthropology”, in which he aims to unravel the ambiguity around the term ‘cloud’ and chart its development from its emergence 20 years ago to what it represents today.
The dawn of public cloud
The film opens with something approaching a montage of experts charting the emergence of public cloud, starting with local data centers in the mid-90s and then moving onto the emergence of what are now described as hyperscalers, and the rush to cloud that followed.
“I think AWS was really the first one to give the services that they developed themselves to others and this is when public cloud became something that was available,” explains Luc Julia, chief scientific officer at car manufacturer Renault Group.
He’s followed by Francesco Bonfiglio, CEO of European federated data infrastructure provider Gaia-X, and Scott Robertson, principal cloud architect at Co-op, remembering the excitement and hype around public cloud at this time.
The pressure to move most or all of your data to the public cloud not necessarily because it was a good solution for your business but because it was what everyone else was doing, as described by both Bonfiglio and Robertson, will doubtless be familiar to many viewers who were also around at this time.
In the words of Robertson, everyone was rushing to join a queue without really understanding why they were doing it. To make matters worse, the end of that queue wasn’t the lovely, fluffy candy floss public cloud that many had hoped for, but was instead something much stickier and a bit less appetizing.
Economics and efficiency
The two flies in the cloud ointment laid out in Clouded are issues of economics and efficiency.
Nominally, access to public cloud infrastructure allows businesses to be more agile and efficient in their operations, particularly in relation to their IT department. It means they don’t have to manage a data center and all the headaches that come with that. This is why Julia became a public cloud fan - for him, public cloud was an exciting and practical solution to his needs, or so he thought.
“Little by little, I realized at the end of the 2010 years – so not long ago – that maybe it’s not the most efficient.”
The reality, according to the interviewees, is that it can be much harder to migrate workloads to the cloud than first thought and if you succeed, it may not work as well. For example, legacy workloads were engineered to work on on-premises hardware, in a private data center. If you migrate them to the cloud, you may find they are slower, less responsive, or otherwise not working at their best.
As Robertson points out, if you’re an ecommerce company you may respond to this by rearchitecting the key application that keeps your business running. For anything else, it may be hard to find the time or resources to fix the issue, and doing nothing will ultimately gnaw away at overall efficiency.
It’s not just legacy apps that are at risk, either. As Kirk Bresniker, chief architect and HPE fellow at Hewlett Packard Labs explains, even things created directly in the cloud can be less efficient than they would have been in the old world, too.
“We’ve taken the meter and we’ve slammed it all the way over to agile; it’s so easy to express a thought and get infrastructure back quickly and easily, but you wonder ‘If I had designed a purpose-built system, how much more efficient would it have been? How fewer bytes would have crossed the network? How fewer servers would have been spun up and spun down? How fewer cycles would it have been if I was back in the old days, when I had a limit on my resources?’,” he says.
It’s not just that public cloud can be less efficient, though: It can also be more costly.
Bill Roth, former director of cloud economics at VMware, says that one of the key premises of cloud is that it’s cheap because you only pay for what you use, and can turn it on when you need it and off when you don’t. The issue with full public cloud, according to Roth, is that about 80% of enterprise applications are on all the time.
“If you buy at the tier where you can turn it on and turn it off, you’re actually spending way too much money,” he says.
Incremental pricing can also make things look deceptively low cost.
“It looks very cheap when you [have] only a fraction of a cent that’s going to be for storage or for sending data, but when you make a list at the end of the month, you see it … can cost a lot,” says Julia.
Data management, data control
As Tremayne-Pitter points out, the popularity of cloud isn’t just about apparent low cost or efficiency – it’s also about data. The public cloud has provided a location for organizations of all types to store vast amounts of data, more of which is being created all the time.
But the ability to store everything is leading to profligacy and laziness; rather than sorting through data and considering what’s useful and what’s not, information is instead being hoarded just in case it’s useful someday. This attitude incurs unnecessary costs and, depending on the type of data and how it’s stored, can create unnecessary risks.
Grant Challenger, director of edge computing at VMware, says that in reality, data has a sell by date in terms of usefulness. “The greater question is ‘where’s the value in this data?’,” he says. “I fundamentally think that the data has value for a certain period of time and then whatever that data is will be reproduced somehow and we’ll have new ways of analyzing it.”
As Julia and Bresniker point out, historical data is a product of its time and having it metaphorically frozen in amber while the world moves on can make it less and less valuable. The passage of time can render this data meaningless and, ultimately, useless, or worse, introduce historical biases that damage present day decision making.
And that’s before questions of sovereignty and control are considered – ensuring that the retention of personally identifiable information (PII) and other sensitive data are stored in accordance with legislation, and that organizations have some transparency and agency over where the location of the hardware on which it resides.
The cloud at the end of the tunnel
For the experts Tremayne-Pitter speaks to, there are various answers to these problems – some hypothetical, some that exist already – but they all agree on the importance of hybrid cloud as a foundation.
Joe Baguley, VP and CTO EMEA at VMware, describes hybrid as “a default”, adding that “most organizations you talk to are already in a hybrid state” with some workloads running in a public cloud, while others are stored on premises or in a colocation provider.
Thomas Maurer, chief evangelist for Azure Hybrid at Microsoft, envisions an even more joined-up future where the cloud will extend from the cloud vendors’ data centers to those of their customers and then on to edge locations.
“Definitely the biggest disruption after the internet is the real distributed, federated cloud era” says Bonfiglio. “In this new era, of course the cloud must be hybrid, of course we’re going to have different types of clouds. The edge requires distributed data centers, distributed cloud and they have to be fit for the purpose.”
And as for Tremayne-Pitter, his journey of cloud anthropology has left him “excited to see how we respond to the growing demands for cloud computing”.
“Hybrid is logical and the promise of the edge is tantalizing. But it’s also safe to say that cloud is no longer a destination, and that cloud first is a retiring trend,” he says, before promising to return soon with further investigation into the nuances of cloud.
You can watch the full Clouded documentary on the Consciously Hybrid site.
