I read somewhere that they transcode everything on the fly, which they insist is necessary. The real point though, imo, is that cloud services have to be incredibly energy intensive in order to justify their existence to begin with.
I don’t think so. Afaik, factors such as subtitles, different languages, different client hardware, mean that transcoding everything on the fly isn’t quite as crazy as you’d first think. I imagine there’s some sort of DRM stuff too, which is going to take its toll.
But I stand by what I said about the business needing the energy cost in order to justify its existence. It’s not just a question of revenue/expenditure — e.g. constantly needing to expand capacity makes a compelling story for investors. Capitalist efficiency, innit
I’m not an expert in this field ofc, but I suspect simply serving a file would be way less energy intensive. There are less centralized alternatives too such as torrent streaming, which may or may not be more efficient. It would be nice to exist in an economy where we could explore these questions!
Just because an economy of scale is real, doesn’t mean the work being done is meaningful or necessary. I’m arguing that the last couple of decades have seen a lot of work being created in order to necessitate traditional ‘economy of scale’ business models — aka a factory with an owner — when other ways of doing things may have been better in terms of global energy efficiency. E.g. the transcoding/compression only needs to happen once for each use case, the whole movie could be buffered rather than maintaining a server connection for the entire runtime. There are examples outside of streaming too ofc, and I’m not saying cloud computing has no use cases — but nobody really believes that the Netflix model is based on sound fundamentals, do they
I think we agree that it’s a both/and situation with economies of scale, so let’s explore Netflix.
The vast majority of energy for Netflix is not in centralized data centers but at their massive number of distribution points. Those distribution points serve a specific problem: end-user-acceptably fast download speeds.
Is that a reasonable problem to solve? Yes, sort of. Instead of people going back to a central server for their content, they go to a regional server. In essence, while end users don’t own their own storage and search infrastructure, the next best solution is a localish solution for the customers in that region.
But what if we just solved the lack of storage and search at home? No profit. So you’ve got that part right for sure
I read somewhere that they transcode everything on the fly, which they insist is necessary. The real point though, imo, is that cloud services have to be incredibly energy intensive in order to justify their existence to begin with.
Oh that could be the explanation.
I wonder if the journalists are lying though. It’s not even in Netflix’s corporate interest to use big energy.
I don’t think so. Afaik, factors such as subtitles, different languages, different client hardware, mean that transcoding everything on the fly isn’t quite as crazy as you’d first think. I imagine there’s some sort of DRM stuff too, which is going to take its toll.
But I stand by what I said about the business needing the energy cost in order to justify its existence. It’s not just a question of revenue/expenditure — e.g. constantly needing to expand capacity makes a compelling story for investors. Capitalist efficiency, innit
I think you’ve got it backwards. The energy cost is real, so economies of scale make sense.
I’m not an expert in this field ofc, but I suspect simply serving a file would be way less energy intensive. There are less centralized alternatives too such as torrent streaming, which may or may not be more efficient. It would be nice to exist in an economy where we could explore these questions!
Than encrypting and decrypting for DRM reasons? Yes. But the reason the cloud exists at all is because economies of scale are real.
Just because an economy of scale is real, doesn’t mean the work being done is meaningful or necessary. I’m arguing that the last couple of decades have seen a lot of work being created in order to necessitate traditional ‘economy of scale’ business models — aka a factory with an owner — when other ways of doing things may have been better in terms of global energy efficiency. E.g. the transcoding/compression only needs to happen once for each use case, the whole movie could be buffered rather than maintaining a server connection for the entire runtime. There are examples outside of streaming too ofc, and I’m not saying cloud computing has no use cases — but nobody really believes that the Netflix model is based on sound fundamentals, do they
I think we agree that it’s a both/and situation with economies of scale, so let’s explore Netflix.
The vast majority of energy for Netflix is not in centralized data centers but at their massive number of distribution points. Those distribution points serve a specific problem: end-user-acceptably fast download speeds.
Is that a reasonable problem to solve? Yes, sort of. Instead of people going back to a central server for their content, they go to a regional server. In essence, while end users don’t own their own storage and search infrastructure, the next best solution is a localish solution for the customers in that region.
But what if we just solved the lack of storage and search at home? No profit. So you’ve got that part right for sure