Distributed computing and p2p business model
Some time ago, while re-reading the LHC wiki page, I happily discovered that CERN had opened its distributed computing project. Joyfully - because the work of the collider has long been very encouraging to me, and the opportunity to somehow help them filled me with stern optimism. I do not know why, but to some extent I naively hope that the result of the project’s work will make our life much better.
Once, my former colleague told me about the SETI @ home project - and therefore I was familiar with the general mechanism of the distributed computing process as a whole. I somehow didn’t really believe in the future of the SETI project, so I didn’t connect to the process. With the LHC, of ​​course, everything turned out differently: BOINC instantly found himself on a computer and, like a pioneer, humbly prepared to take part in the solution of the structure of the universe. But, to my deep disappointment, there were no assignments for calculation and it was not foreseen.
To keep my enthusiasm dry, I added a couple of projects to combat various diseases, I curtailed the manager and began reading all sorts of articles about distributed computing. At the same time, the thought flashed through my head about how similar it was to p2p. And here I had an idea, which I would like to share, or rather, find out your professional opinion about the possibility / impossibility of its appearance in life, and perhaps about the existence of realizations and their successes / failures.
This idea is contained in the following business model: there is a) a customer who needs either some kind of file storage with a wide distribution channel, or some computational power, or both in aggregate b) a crowd of customers with their idle home / work / mobile computers on which a quietly running background client is installed within strictly defined limits of resources taken in) the intermediary who organizes the work of the first with the second. The customer pays a certain money for services: 100% for a certain amount of services (storage or payments), the intermediary takes 5% of them for food, bread and process organization, and the remaining 95% is distributed to those customers who directly perform the task: file sharing and / or calculation.
')
It is important to note here that file storage does not imply, say, storage of a database file. Of course, the reaction rate of such a base would be unacceptable. We are talking about, say, the storage of legal software online store (Steam, AppStore, OviStore). Or storage backups of large services. Or video hosting. Those. in file storage, high distribution speed (plus reliable storage) is important, not pinging data returns.
Calculations can be useful in various areas (although there is a problem with converting data into a suitable format, but it seems to me that it can be solved as a whole). For example, the Avatar film was rendered on a cluster, but it could well be handled by such a network, and perhaps it would have come out much faster and cheaper.
In theory, everyone will be satisfied: the customer is on the issue price, and the intermediary and clients are from a certain income. Plus, this is a huge impetus for the market, which will have the opportunity to use relatively inexpensive, but extremely powerful resources for computing and high-speed storage. It may not turn out all at once, but potentially - it can grow into a large-scale network: any file from the Internet will download very quickly, but at its own penny from other users, and such a system, theoretically, can balance itself - as much as I download, distribute it as much what speed I download, with this and distribute, otherwise I either earn or spend.
I understand that questions arise the sea. For example, the question of the legality of stored files - but here you can apply the same schemes as when stored in file storages. The main thing is to make sure that the data is encrypted and it was impossible to a) steal other people's data (use splitting, certificates, keys) b) the customer or intermediary could always delete or block files (say, through certain revision files distributed along with chunks of files) .
The issue of high power in bad hands can be compensated by the degree of verification depending on the required. If someone requires sufficient power to refactor keys, he must somehow justify his calculation (at least identify himself as a company with certain minimum requirements). In any case, there are already millions of zombie computers in the world, which in the hands of hackers probably already perform similar tasks, and if such a global danger existed - it would have already “exploded”. With a distributed network, it would be possible to provide similar power, but for useful business tasks.
The same question arises: is it worth it - is it not easier to build an additional powerful data center? But he will demand: the cost of carriers, the cost of rent, the remuneration of employees, the cost of electricity, the cost of the channel. Plus depreciation: the physical equipment needs to be updated periodically, users update their computers themselves and there is no need to pay for it. Plus reliability: the data center can burn and the distributed network will survive. Scalability: today you need 50% of the power of the data center, tomorrow 100%, the day after tomorrow 150%, and after-after tomorrow - again 50%. Rent, wages, etc. - all these are static amounts that can easily turn into losses. The network of users will be much more flexible to respond to the market according to the classical supply and demand scheme (I do not have enough earnings at 50 rubles a month - I will not participate and leave, the rest will receive 51 rubles a month, and vice versa).
Would such participation be economically interesting to the clients? You can try to estimate: my home computer can easily give 10GB per day without interfering with my work, which means it is about 300GB per month. At the same time, in order to quickly download the size file I need in 10GB, in principle, I am ready to pay 50p. and, it seems to me, I am not the only one (otherwise there would be no “vip” access on file hosting services). Those. in case of full load it is about 1500r. (minus 5% to the intermediary) per month - the canal and the new screw will pay off in six months or a year. If the network does not need to be 300GB per month - I will only give, say, 10GB and earn 50p. - even such a meager amount will be pleasant for, by and large, the minimum cost of the channel and electricity, while my computer is still turned on and running.
At the same time, something tells me that if such a business model had the right to exist, it would already exist. What prevents her from being born?
Once, my former colleague told me about the SETI @ home project - and therefore I was familiar with the general mechanism of the distributed computing process as a whole. I somehow didn’t really believe in the future of the SETI project, so I didn’t connect to the process. With the LHC, of ​​course, everything turned out differently: BOINC instantly found himself on a computer and, like a pioneer, humbly prepared to take part in the solution of the structure of the universe. But, to my deep disappointment, there were no assignments for calculation and it was not foreseen.
To keep my enthusiasm dry, I added a couple of projects to combat various diseases, I curtailed the manager and began reading all sorts of articles about distributed computing. At the same time, the thought flashed through my head about how similar it was to p2p. And here I had an idea, which I would like to share, or rather, find out your professional opinion about the possibility / impossibility of its appearance in life, and perhaps about the existence of realizations and their successes / failures.
This idea is contained in the following business model: there is a) a customer who needs either some kind of file storage with a wide distribution channel, or some computational power, or both in aggregate b) a crowd of customers with their idle home / work / mobile computers on which a quietly running background client is installed within strictly defined limits of resources taken in) the intermediary who organizes the work of the first with the second. The customer pays a certain money for services: 100% for a certain amount of services (storage or payments), the intermediary takes 5% of them for food, bread and process organization, and the remaining 95% is distributed to those customers who directly perform the task: file sharing and / or calculation.
')
It is important to note here that file storage does not imply, say, storage of a database file. Of course, the reaction rate of such a base would be unacceptable. We are talking about, say, the storage of legal software online store (Steam, AppStore, OviStore). Or storage backups of large services. Or video hosting. Those. in file storage, high distribution speed (plus reliable storage) is important, not pinging data returns.
Calculations can be useful in various areas (although there is a problem with converting data into a suitable format, but it seems to me that it can be solved as a whole). For example, the Avatar film was rendered on a cluster, but it could well be handled by such a network, and perhaps it would have come out much faster and cheaper.
In theory, everyone will be satisfied: the customer is on the issue price, and the intermediary and clients are from a certain income. Plus, this is a huge impetus for the market, which will have the opportunity to use relatively inexpensive, but extremely powerful resources for computing and high-speed storage. It may not turn out all at once, but potentially - it can grow into a large-scale network: any file from the Internet will download very quickly, but at its own penny from other users, and such a system, theoretically, can balance itself - as much as I download, distribute it as much what speed I download, with this and distribute, otherwise I either earn or spend.
I understand that questions arise the sea. For example, the question of the legality of stored files - but here you can apply the same schemes as when stored in file storages. The main thing is to make sure that the data is encrypted and it was impossible to a) steal other people's data (use splitting, certificates, keys) b) the customer or intermediary could always delete or block files (say, through certain revision files distributed along with chunks of files) .
The issue of high power in bad hands can be compensated by the degree of verification depending on the required. If someone requires sufficient power to refactor keys, he must somehow justify his calculation (at least identify himself as a company with certain minimum requirements). In any case, there are already millions of zombie computers in the world, which in the hands of hackers probably already perform similar tasks, and if such a global danger existed - it would have already “exploded”. With a distributed network, it would be possible to provide similar power, but for useful business tasks.
The same question arises: is it worth it - is it not easier to build an additional powerful data center? But he will demand: the cost of carriers, the cost of rent, the remuneration of employees, the cost of electricity, the cost of the channel. Plus depreciation: the physical equipment needs to be updated periodically, users update their computers themselves and there is no need to pay for it. Plus reliability: the data center can burn and the distributed network will survive. Scalability: today you need 50% of the power of the data center, tomorrow 100%, the day after tomorrow 150%, and after-after tomorrow - again 50%. Rent, wages, etc. - all these are static amounts that can easily turn into losses. The network of users will be much more flexible to respond to the market according to the classical supply and demand scheme (I do not have enough earnings at 50 rubles a month - I will not participate and leave, the rest will receive 51 rubles a month, and vice versa).
Would such participation be economically interesting to the clients? You can try to estimate: my home computer can easily give 10GB per day without interfering with my work, which means it is about 300GB per month. At the same time, in order to quickly download the size file I need in 10GB, in principle, I am ready to pay 50p. and, it seems to me, I am not the only one (otherwise there would be no “vip” access on file hosting services). Those. in case of full load it is about 1500r. (minus 5% to the intermediary) per month - the canal and the new screw will pay off in six months or a year. If the network does not need to be 300GB per month - I will only give, say, 10GB and earn 50p. - even such a meager amount will be pleasant for, by and large, the minimum cost of the channel and electricity, while my computer is still turned on and running.
At the same time, something tells me that if such a business model had the right to exist, it would already exist. What prevents her from being born?
Source: https://habr.com/ru/post/96096/
All Articles