I am reminded of the theory of the firm; in brief: the transaction costs of a free market can easily exceed the value of competition. A “serverless” model is full of transaction costs – during development, learning or building all the APIs, during runtime network latency and bandwidth, and during maintenance, debugging costs.
Couple that with the observation that much “big data” is not actually that big, and will fit into RAM on a single non-special-purpose server.
Also coming to mind: “trusted system” doesn’t mean you believe it to be reliable or secure, it means that you have it as a dependency.
more of passing thought that than a nuanced response, but I find it disheartening that pieces like this completely fail to take into account or even consider the underlying power structures (basically politics at various levels) that have bothed massively influenced what “cloud” we have today, where it can realistically go tomorrow, and what the implications of any predicited routes might be, (e.g is this further centralising control, how is the balance of power being changed between various cloud providers, device manufacturers, network owners, etc etc)
It’s not so much conscious political decisions as blind corporate seeking of efficiencies.
as maybe, but I would hope that even in the most boring way possible (i.e who is the big fish in my pond) at least some corporate entities are aware of the wider political implications of where cloud computing might go and what it might do to them. Just chasing efficiency by itself is not a viable long term stratergy
Yes, but you can also substitute “externalities” for “efficiencies” most of the time.