4 ms·

I have seen all sorts of different database technologies being touted as the "next generation" database technology over 4 decades. Each has not superseded any o

I have seen all sorts of different database technologies being touted as the "next generation" database technology over 4 decades. Each has not superseded any of the others. What ends up happening is that a "new" database technology finds an application area for which it is the "best fit" and is then used accordingly. Some of theses application areas can be broad and some can be very specialised. If you can build a usable system and can make a market for it, do so. If you think it has broad application and want it to be ubiquitous with a variety of players (including FOSS systems) providing implementations then release all the relevant implementation and research information. But you first have to decide if you want to give the basic ideas away or whether you want to hold them close to your chest. Until you make that decision, a lot of the questions you have asked are not going to be so answerable. There are consequences for giving the information away and there are consequences for keeping it close to your chest (including patents, trade secrets, etc.) The application fields where you think you technology may be benefited from would require you actually building full-blown demonstrations systems against which you can compare existing technologies. I have seen many different systems over the decades which were simply "flash in the pan" technologies and as such take your comments with a couple of tonnes of sea-salt. If your technology is both viable and worth the price and convincing, you might make it a successful venture.

craxel 7y ago
Thank you for taking the time to reply. We have so far kept things proprietary and we do have patents. We've done several proof of concepts with very large companies and they've been successful and moving forward. We have a "partnership" with a very recognizable IT company that reviewed our technology. However, the sales cycle is really long and a lot of "that's too good to be true" - so thinking about the ubiquitous path. I started programming in 1982 - so I've seen a lot and understand the skepticism. However, the choices made in the 70's about indexing (BTrees) aren't the best choice for a highly parallel, distributed world. Plus BTrees order the data, so they'll likely never work for "secret" indexing. For instance, it is quite amusing - all of the interest in fully homomorphic encryption and it being the holy grail if they can just make it fast enough. FHE is a secure computation technology, not a scalable search technology. Let's say you have 1 billion records - with FHE, you'd have to evaluate all 1 billion records for every query (which would take days at best current rate of 1 million evals per hour). As soon as you index those 1 billion records to avoid that with a BTree, you are subject to substantial inference risk because the records would be ordered. So, without an index that has limited inference risk, FHE is DOA for indexing/search (it is too slow regardless). If you have that index with limited inference risk, you don't need FHE for indexing/search. I would argue that our data structures for indexing data are ready for an update - in a world with exploding volume and velocity of data and huge security and privacy risk. Think about it, one hacker breaches a database server today and they get everything. If we keep concentrating valuable info like this, it'll keep getting stolen. Again, appreciate your reply.
- oldandtired 7y ago
  In regards to "gackers", "crackers" and "mackers" devising ways to access the data, these kinds of people will always be a hazard to deal with. There are all sorts of ways to deal with this, including not actually collecting all such data in the first place, but that's a discussion for another time and place. As far as having proprietary model and the use of patents, these already dictate how you will move forward from here and as such, you now have to have the do the hard yards to bring everything to fruition. You have to ask yourself this question: do you have the necessary nouse to follow the business model through to completion? Just remember that when you bring in business "partners", they will have their own priorities that will concern them and yours will be secondary or even tertiary to them. You have to stay ahead of the games that you are involved with (there will be many of these) and it is very easy to lose sight of this and drop the ball. Finally, just remember that there are no guarantees that even if you have the best technology and all the patents that surround this new technology that there will be anyone who will be interested in what you have to offer at the price you want to sell it at. From a personal perspective, I no longer have any interest in making anything proprietary or ever getting in the ugly world of patents. So take my comments above in that light. I have run my own business and it is a lot of hard work to keep it going and growing. I no longer have any such interest as I find myself now in a place of having far more important things to interested in and devote my time to.

craxel 7y ago

Thank you for taking the time to reply. We have so far kept things proprietary and we do have patents. We've done several proof of concepts with very large companies and they've been successful and moving forward. We have a "partnership" with a very recognizable IT company that reviewed our technology. However, the sales cycle is really long and a lot of "that's too good to be true" - so thinking about the ubiquitous path. I started programming in 1982 - so I've seen a lot and understand the skepticism. However, the choices made in the 70's about indexing (BTrees) aren't the best choice for a highly parallel, distributed world. Plus BTrees order the data, so they'll likely never work for "secret" indexing. For instance, it is quite amusing - all of the interest in fully homomorphic encryption and it being the holy grail if they can just make it fast enough. FHE is a secure computation technology, not a scalable search technology. Let's say you have 1 billion records - with FHE, you'd have to evaluate all 1 billion records for every query (which would take days at best current rate of 1 million evals per hour). As soon as you index those 1 billion records to avoid that with a BTree, you are subject to substantial inference risk because the records would be ordered. So, without an index that has limited inference risk, FHE is DOA for indexing/search (it is too slow regardless). If you have that index with limited inference risk, you don't need FHE for indexing/search. I would argue that our data structures for indexing data are ready for an update - in a world with exploding volume and velocity of data and huge security and privacy risk. Think about it, one hacker breaches a database server today and they get everything. If we keep concentrating valuable info like this, it'll keep getting stolen. Again, appreciate your reply.

oldandtired 7y ago
In regards to "gackers", "crackers" and "mackers" devising ways to access the data, these kinds of people will always be a hazard to deal with. There are all sorts of ways to deal with this, including not actually collecting all such data in the first place, but that's a discussion for another time and place. As far as having proprietary model and the use of patents, these already dictate how you will move forward from here and as such, you now have to have the do the hard yards to bring everything to fruition. You have to ask yourself this question: do you have the necessary nouse to follow the business model through to completion? Just remember that when you bring in business "partners", they will have their own priorities that will concern them and yours will be secondary or even tertiary to them. You have to stay ahead of the games that you are involved with (there will be many of these) and it is very easy to lose sight of this and drop the ball. Finally, just remember that there are no guarantees that even if you have the best technology and all the patents that surround this new technology that there will be anyone who will be interested in what you have to offer at the price you want to sell it at. From a personal perspective, I no longer have any interest in making anything proprietary or ever getting in the ugly world of patents. So take my comments above in that light. I have run my own business and it is a lot of hard work to keep it going and growing. I no longer have any such interest as I find myself now in a place of having far more important things to interested in and devote my time to.