11 ms·
I have never understood why RDBs have such general concepts of primary keys, where you can for example let date of birth be a primary key, when in every schema
by whatshisface 5y ago
I have never understood why RDBs have such general concepts of primary keys, where you can for example let date of birth be a primary key, when in every schema I have ever designed or seen, all rows get unique integer IDs anyway.
- mmcgaha 5y agoI completely agree with you, yet I design most of my schemas with integer keys anyway. The only thing I do different than most folks is that I make a unique constraint/index for candidate keys (the proper primary key) because otherwise we have lost the ability to ensure data consistency. As for why I use integer keys, it has always been the convention everywhere I go so I don't want to break convention. Now the only convention that I would rather see would be to use GUIDs instead of integers because I hate that I can join an item table to an employee table and get results back.
- cyberbanjo 5y agoWhat about people with the same birthdays?
- wfriesen 5y agoBusiness/requirements/users will often tell you that no two users will have the same birthday/SSN/whatever, but so often that eventually proves incorrect. Just using an auto increment ID as the primary key can prevent a lot of pain down the road.
- lifepillar 5y agoSo, when someone comes to you and identifies herself by providing her SSN, name, date of birth, how do you know whether she corresponds to the record with id M or to the record with id N? Or, more likely, to one of the N records, with N>2, with that SSN, name, date of birth?
- jraph 5y agoWhat about them? Do they really exist, practically speaking? At a quick glance on my databases, I see nothing of the sort.
- lkuty 5y agohttps://en.wikipedia.org/wiki/Birthday_problem https://en.wikipedia.org/wiki/Birthday_problem
- jraph 5y agoIs there anything comparable to the birthday paradox, which includes years, by the way? I would guess that you would need a big group of random people such that there is a high chance of having two people born the very same day.
- whatshisface 5y agoThe birthday paradox threshold if years are included would be 23 people per year, rather than 23 people in your database. What that translates to in your database depends a lot the age demography of your userbase. If people live to be 100, and if I don't worry about the fact that some years will have more people than others even if they are randomly distributed (something that should go in to the model), that's about 2,300 people before you start expecting to see duplicate birthdays.
- cyberbanjo 5y agoI don't think the birthday problem accounts for year, but yet it's clear some people share the same birthdate, twins for example.
- jcranmer 5y agoI share the same name and birthdate as another individual. It has actually caused me to be held up at the border before.
- jraph 5y ago
- ARandomerDude 5y agoI think this was the parent's point. Since, as you point out, we we almost never use the data itself (e.g. date of birth) as the primary key, the parent is asking why is this a feature at all? Why not just mandate a uuid, number, etc. as the primary key?
- uDontKnowMe 5y agoWell it's easy enough to think of a situation where you would use a date as a primary key, for example time series data.
- turbocon 5y agoIt's rare in practice but occasionally rows have natural unique identifiers that are useable. The textbook case would be student IDs in a university database where each row corresponds to a single student. However some universities view the student id as sensitive information (kind of like a ssn in the US) and so in that scenario it should be aliased to some other key to prevent the student id from being present in multiple tables as a foreign key.
- TeMPOraL 5y ago> However some universities view the student id as sensitive information (kind of like a ssn in the US) and so in that scenario it should be aliased to some other key to prevent the student id from being present in multiple tables as a foreign key. It is ironic, given that the whole reason to have a student ID is for it to be a primary key in some database somewhere.
- whatshisface 5y agoDatabases are vulnerable to a particular type of folkloric magic: https://en.m.wikipedia.org/wiki/True_name https://en.m.wikipedia.org/wiki/True_name
- jameshart 5y agoStudent IDs are certainly not ‘natural’ keys. I am almost entirely convinced that there is actually no such thing as a natural key.
- mulmen 5y agoNatural keys are everywhere. They are just less interesting than the relationships. My name is a natural key. It uniquely identifies my name. The wheels come off if you expect it to uniquely identify me. But with natural keys that's never actually necessary. When I buy a car they record my name, my date of birth, the time of the sale and some details about the car, such as the VIN, which is its name. That provides enough information to form a natural key to identify the sale. At no point does a database actually have a concept of who I am really, only the relevant data points to satisfy its own relational model. Any database that means to track information about people would be able to construct a natural key from my name, my mothers name and some information about my birth, such as the doctor, hospital and time.
- dnlhg 5y agoCodd was pretty adamant in his original papers that keys should be made up from data rather than a unique integer, and his example of converting from a hierarchical database schemas to a relational database schema shows how compound, natural keys are important to represent relationships. But we don’t usually do it that way today, partly because it isn’t that practical with our current databases and ORMs. It has some theoretical advantages and disadvantages, but using IDs seems to work well enough that we’re not investing in a new paradigm.
- wwweston 5y agoYeah, I think the early proponents of relational models had a lot to do with setting the culture here. I suspect that some of this has to do with an insight I recently discovered: a "natural" key is in some sense a key that's so foreign it's in a "database" you don't control. And if you establish a culture that converges on natural keys, you may end up with more easily cross-operable databases/collections as an asset. And this is why I think it's probably dead. If the boring practicality of generated identifiers didn't do it, laws like the CCPA which are going to, because they recognized the potential of broad cross-operable data to be a liability and a hazard vs an asset.
- whatshisface 5y agoThose "natural" databases as a rule are going to be full of duplicates and sometimes even inconsistencies, hence the need for IDs that only exist for the purpose of expressing relationships within your own database. Basically, pointers.
- CRConrad 5y agohttps://news.ycombinator.com/item?id=27658993 https://news.ycombinator.com/item?id=27658993
- ARandumGuy 5y agoIn practice, using a unique IDs is often not ideal, but it also rarely the wrong choice. I've personally never heard of a scenario where a unique ID caused difficult to fix issues (although I'm sure it's possible). In contrast, a poorly though out natural key can cause a lot of headaches. This mostly occurs when your natural key isn't as unique as you thought, or if you have to change what you use as a natural key. The real world does not like to conform to our database constraints.
- dylan604 5y agoHow would DoB work as a primary key? A primary key is useless if not unique, right? And I know many people born on the same day that aren't even twins. Incrementing integers are nice and easy for humans to consume. Using GUIDs is far more portable than unique integers, but damn near impossible for humans to deal.
- whatshisface 5y agoThey wouldn't, so my point is why do RDBMSs support so many kinds of primary keys?
- dylan604 5y agoKeys can span multiple columns, so DoB could be a differentiator if First, Last are also the same. I personally would never try to build a key like that though, but that's the only way I could see justifying its use. If you can't see a why for uses of a field that is truly more unique, then I'd suggest you just need some more imagination. Strictly not allowing it seems to me to be short sighted just because someone had the ill conceived idea to use DoB as a primary key
- whatshisface 5y agoCan you think of any good use case for a non-arbitrary primary key? You are right that first name + last name + dob is not a good way to design the system. For example someone could differentiate their twin children only by their middle name. So, what's an example of a non-integer non-arbitrary primary key?
- dylan604 5y agoNot sure exactly what you mean by non-arbitrary primary key, but... GUID/UUID FeatureName/ReleaseYear or Series/Season/Episode email phone-number SSN driverslicense#/State
- bart_spoon 5y agoI stated this elsewhere, but with some sports data, you don’t gain anything with a uniquely generated game ID. NBA teams will only play one game at most per day, and only play one other team, so if you make a key of the date and the teams playing, you uniquely identify every event. You can create an arbitrary ID for a game if you want, but it’s not actually giving you information you don’t have, and therefore is entirely redundant.
- mamcx 5y ago> when in every schema I have ever designed or seen, all rows get unique integer IDs anyway. This is artifact of 4 things: - Indexed Integer search are fast in RDBMS - Having an non-bussines related value for identify a row could save you from making triggers for updated in other relations if it change. This is the part a lot of folks missed: If you already have a natural PK and it not change, is wasteful add ANOTHER (is another index btw) - Is enforced by ORMS that are badly designed. - Most people have not idea how model a database, and this one little thing is one of the easier "fix" you can think of! I have done a lot "non integer primary keys" before (and not GUIds!) and is useful that RDBMS are not like mongo with a fixed schema (yes! Mongo is a single fixed-schema for all your data!). For reporting, store aggregates, pre-compute values, etc is very valuable! For example, in accounting you do reports per day/mont/year. You then do granularity per day, and the another per month, per year, per semestre, etc. When the data volume is high, making tables "days, months, years" make a lot of sense. Cool guys call this a "time-series database". But the rdbms CAN model it! And can model - "document database" - "key-value database" - "olap database" - "columnar database" etc. (you go for specialized backend for performance but honestly? Most of the time is a mistake if them are not relational. "Relational" don't means "is backed by b-trees, is row oriented and use SQL")
- whatshisface 5y agoA time series database isn't keyed by day/month/year, it's typically keyed by monotonic timestamps.
- mamcx 5y agoYes, that is even a more narrowing detail. (people that use a "time-series" rarely need that high of granularity!) But the question was about wondering "why exist the support for PK that are not integers/guis, like dates"?. To be able to model a lot of things for what in the past (before RDBMS) and today (for upstarts engines) you will be tempted to add another data store (that rarely is that necessary). I know by a fact that some times is because people not realize you DON'T need to constrain your db to things like "only integers pks!"! P.D: I work in the enterprise sector. Is fun when something is composed of a cluster of disparate things purely by misunderstanding you could have solved it with a simple db, well done. Contrary to belief, a lot of my customers, their largest databases are 1-10GB each. A rdbms can happily go much much larger than that, if you give them a little love...
- hknapp 5y agoI worked at a place that used decimals that auto incremented by 2
- kbelder 5y agoSmart. That allows them to insert new rows without messing up the ordering.
- dragonwriter 5y ago> I have never understood why RDBs have such general concepts of primary keys, where you can for example let date of birth be a primary key, when in every schema I have ever designed or seen, all rows get unique integer IDs anyway. Because sequential integer surrogate keys aren't always a good idea, e.g.: 1. Certain logically-unique join tables, where the foreign keys to the joined tables form an obvious composite primary key. 2. Cases with natural primary keys for basic data; A date might not be a good primary key if its the DoB in a table of people, but its an excellent primary key in a calendar table. 3. Even when you need surrogate keys, sometimes you need them to be able to be generated in a distributed manner, so you want something like UUID/ULID instead of sequential integers.
- projektfu 5y agoIn one sense, the unique integer is often not real data. It’s synthetic and has no meaning outside the application. I find it easier to understand when I think that Codd and other database pioneers were looking at creating universal databases, not the application-specific ones we tend to use. Furthermore, the use of auto numbering tends to make the queries easier to write but the tables less meaningful. For example, INVOICE might have a sequential invoice id that is real information, as well as a reference to customer, billing address, shipping address, etc. INVOICE_ITEM will have a foreign key to an invoice, unique number, position, stock item id (second foreign key), description, etc. PACKING_LIST might have a foreign key to INVOICE_ITEM. In my opinion, I prefer to see INVOICE_ID as part of the primary key of INVOICE_ITEM. I like being able to see the relationship in foreign keys and things like that. These things can be a heated topic and other designers probably think I also like to murder kittens. But the data layout is, I think, what Codd was going for.
- fuckf4ce 5y agoAvoiding "natural keys" has little to do with relational databases or the model. Consider that the Social Security Number far predates relational databases. The reason for using surrogate keys is not some great conspiracy or mass incompetence. Simply, there are very few cases where natural keys work well. Good keys are generally stable, globally unique, not too big, and it turns out that these qualities are extremely rare in a natural key in practice. Furthermore, composite keys over more than 2 or 3 attributes become terribly unwieldy (and intuitively, composite key size and stability are inversely related). What is a good natural key for an individual? There are none! The SSA figured this out in the 20s, this was not a modern tech problem. Even social security numbers can't be stable, but they are far better than things like names or addresses because they change for only a handful of reasons. That said, a lot of database designs should use composite primary keys. You can in part thank early MySQL which started the ball rolling on generally shitty database design in the late 90s.
- brazzy 5y agoSince when is "date of birth" a distinct type? And "date" definitely does make sense as a primary key when you have a table that has conceptually at most one entry per day. For example, a travel journal. Or a time sheet application.
- juped 5y agoSynthetic primary keys like a unique integer ID assigned to a row are worse in theory - the id column isn't real data. They exist as a very common practical compromise for good reasons, but you should be aware you're compromising theory for practice.
- adrianmonk 5y agoWhat you're talking about is a natural key[1], which can be contrasted with a surrogate key[2]. (Surrogate keys are also called "synthetic keys" or "factless keys".) So the question is whether natural keys are ever useful. I think the answer is yes, they are useful if the real world already imposes a key on the stuff you are modeling in the database. In that case, it is already unique (as it needs to be), and generating your own key doesn't accomplish anything (but is redundant). For example, suppose I'm creating a history database of who owned what internet domain names at which times. If my table is one-to-one with domains, I can use the domain name itself as the primary key. If not, I can perhaps use the domain name plus some other column as a composite key. For a table of ownership intervals, my primary key could be this: domain name + start date + end date. --- [1] https://en.wikipedia.org/wiki/Natural_key https://en.wikipedia.org/wiki/Natural_key [2] https://en.wikipedia.org/wiki/Surrogate_key https://en.wikipedia.org/wiki/Surrogate_key
- whatshisface 5y ago>For example, suppose I'm creating a history database of who owned what internet domain names at which times. If my table is one-to-one with domains, I can use the domain name itself as the primary key. In your example, wouldn't that not work, because the same domain would be in the table twice with two different owners?
- adrianmonk 5y agoYou might have more than one table in the database. For example, you could have a table of current owners for quick lookup and a table of all owners for full history. For the current owners table, you'd use just the domain as a key. For all the full history table, you'd use domain + start date + end date as the key. So yeah, that makes it a weird example, but the point is that in neither table did you make up a synthetic key. In one table, you used a real-world fact (domain) as the key, and in the other table you used three real-world facts (domain plus two dates) together as a key.