Related
To keep this as short as possible I'm going to use and example.
So let's say I have a simple database that has the following tables:
company - ( "idcompany", "name", "createdOn" )
user - ( "iduser", "idcompany", "name", "dob", "createdOn" )
event - ( "idevent", "idcompany", "name", "description", "date", "createdOn" )
Many users can be linked to a single company as well as multiple events and many events can be linked to a single company. All companies, users and events have columns as show above in common. However, what if I wanted to give my customers the ability to add custom fields to both their users and their events for any unique extra information they wish to store. These extra fields would be on a company wide basis, not on a per record basis ( so a company adding a custom field to their users would add it to all of their users not just one specific user ). The custom fields also need to be sesrchable and have the ability to be reported on, ideally automatically with some sort of report wizard. Considering the database is expected to have lots of traffic as well as lots of custom fields, what is the best solution for this?
My current research and findings in possible solutions:
To have generic placeholder columns such as "custom1", "custom2" etc.
** This is not viable as there will eventually be too many custom columns and there will be too many NULL values stored in the database
To have 3x tables per current table. eg: user, user-custom-field, user-custom-field-value. The user table being the same. The user-custom-field table containing the information about the new field such as name, data type etc. And the user-custom-field-value table containing the value for the custom field
** This one is more of a contender if it were not for its complexity and table size implications. I think it will be impossible to avoid a user-custom-field table if I want to automatically report on these fields as I will have to store the information on how to report on these fields here. However, In order to pull almost any data you would have to do a million joins on the user-custom-field-value table as well as the fact that your now storing column data as rows which in a database expected to have a lot of traffic as well as a lot of custom fields would soon cause a problem.
Create a new user and event table for each new company that is added to the system removing the company id from within those tables and instead using it in the table name ( eg user56, 56 being the company id ). Then allowing the user to trigger DB commands that add the new custom columns to the tables giving them the power to decide if it has a default value or auto increments etc.
** Everytime I have seen this solution it has always instantly been shut down by people saying it would be unmanageable as you would eventually get thousands of tables. However nobody really explains what they mean by unmanageable. Firstly as far as my understanding goes, more tables is actually more efficient and produces faster search times as the tables are much smaller. Secondly, yes I understand that making any common table changes would be difficult but all you would have to do is run a script that changes all your tables for each company. Finally I actually see benefits using this method as it would seperate company data making it impossible for one to accidentally access another's data via a potential bug, plus it would potentially give the ability to back up and restore company data individually. If someone could elaborate on why this is perceived as a bad idea It would be appreciated.
Convert fully or partially to a NoSQL database.
** Honestly I have no experience with schemaless databases and don't really know how dynamic user defined fields on a per record basis would work ( although I know it's possible ). If someone could explain the implications of the switch or differences in queries and potential benefits that would be appreciated.
Create a JSON column in each table that requires extra fields. Then add the extra fields into that JSON object.
** The issue I have with this solution is that it is nearly impossible to filter data via the custom columns. You would not be able to report on these columns and until you have received and processed them you don't really know what is in them.
Finally if anyone has a solution not mentioned above or any thoughts or disagreements on any of my notes please tell me as this is all I have been able to find or figure out for myself.
A typical solution is to have a JSON (or XML) column that contains the user-defined fields. This would be an additional column in each table.
This is the most flexible. It allows:
New fields to be created at any time.
No modification to the existing table to do so.
Supports any reasonable type of field, including types not readily available in SQL (i.e. array).
On the downside,
There is no validation of the fields.
Some databases support JSON but do not support indexes on them.
JSON is not "known" to the database for things like foreign key constraints and table definitions.
Imagine you have a SQL database like mysql or postgresql. You have two tables : user and car. One user can drive N cars, a car can be driven by N users, so you have a third "drive" table with two foreign key.
Now, you want that your table user goes on elasticsearch, because you want search users by name, email... etc... Maybe you also need to do some search on the car table.
I see three way to achieve this, I d'like to know what is the best way :
1) Abandon the sql database. All your tables are now on elasticsearch. You can do search on whatever you want, but you must treat all your constraints manually.
2) Keep the structure on the sql database, you keep your three tables, the primary keys and the foreign keys. But your tables contains only elasticsearch ID of the associated row in elasticsearch. For exemple in table user, you keep user_id and add a user_elasticsearch_id that point on the elasticsearch row where you found the name, the email... etc... So you have your sql constraints, you can do search, but you must maintain two tables.
3) Duplicate. You don't touch your sql database, you duplicate all the rows on the elasticsearch database. You have your constrains, you can search, but again you must maintain two tables and you have twice the data and twice the storage.
Now, brave fellow of stackoverflow, what would you do in this case ?
Thank you.
The most common setup for critical business data is having e.g. a SQL database as your primary datastore and Elasticsearch as additional search index. (= your solution 3).
An alternative for non business-critical data like logs etc. is having Elasticsearch standalone.
Solution 2 seems wired, is not an option for me.
Because you may have a lot of business rules mixed into you database and application using it, I would be conservative and keep the DB. And use ES to index the user attributes I want to search on. ES would return scored results. When a result select I would switch to DB to retrieve all information and relations.
So I would choose 2b : keep DB and store PK in ES, not ID in DB).
Keep in mind you can force the ID en ES. It could be "user_PK" or something alike.
So, I've read a lot about how stashing multiple values into one column is a bad idea and violates the first rule of data normalisation (which, surprisingly, is not "Do Not Talk About Data Normalisation") so I need some help.
At the moment I'm designing an ASP .NET webpage for the place I work for. I want to display data on a web page depending on what Active Directory groups the person belongs to. The first way of doing this that comes to mind is to have a table with, essentially, a column containing the AD group and the second column containing what list of computers belong to that list.
I've learnt that this is showing great disregard for relational databases, so what is a better way to do it? I want to control this access by SQL tables, so I can add/remove from these tables and change end users access accordingly.
Thanks for the help! :)
EDIT: To describe exactly what I want to do is this:
We have a certain group of computers that need to be checked up on, however these computers are in physically difficult to reach locations. The organisation I belong to has remote control enabled for these computers, however they're not in the business of giving out the remote control password (understandable).
The added layer of complexity is that, depending on who you are, our clients should only be able to see a certain group of computers (that is, the group of computers that their area owns). So, if Group A has Thomas in it, and Group B has Jones in it, if you belong to either group then you would just see one entry. However, if you belong to both groups you should see both Thomas and Jones computers in it.
The reason why I think that storing this data in a SQL cell is the way to go is because, to store them in tables would require (in my mind) a new table for each new "group" of computers. I don't want to crank out SQL tables for every new group, I'd much rather just have an added row in a SQL table somewhere.
Does this make any sense?
You basically have three options in SQL Server:
Storing the values in a single column.
Storing the values in a junction table.
Storing the values as XML (or as some other structured data format).
(Other databases have other options, such as arrays, nested tables, and JSON.)
In almost all cases, using a junction table is the correct approach. Why? Here are some reasons:
SQL Server has (relatively) lousy string manipulation, so doing something as simple as ensuring a unique list is really, really hard.
A junction table allows you to store lots of other information (When was a machine added? What is the full description of the machine? etc. etc.).
Most queries that you want are pretty easy with a junction table (with the one exception of getting a comma-delimited list, alas -- which is just counterintuitive rather than "hard").
All the types are stored natively.
A junction table allows you to enforce constraints (both check and foreign key) on the elements of the list.
Although a delimited list is almost never the right solution, it is possible to think of cases where it might be useful:
The list doesn't change and presentation of the list is very important.
Space usage is an issue (alas, denormalization often results in fewer pages).
Queries do not really access elements of the list, just the entire thing.
XML is also a reasonable choice under some circumstances. In the most recent versions of SQL Server, this can be made pretty efficient. However, it incurs the overhead of reading and parsing XML -- and things like duplicate elimination are still not obvious.
So, you do have options. In almost all cases, the junction table is the right approach.
There is an "it depends" that you should consider. If the data is never going to be queried (or queried very rarely) storing it as XML or JSON would be perfectly acceptable. Many DBAs would freak out but it is much faster to get the blob of data that you are going to send to the client than to recompose and decompose a set of columns from a secondary table. (There is a reason document and object databases are becoming so popular.)
... though I would ask why are you replicating active directory to your database and how are you planning on keeping these in sync.
I not really a bad idea to store multiple values in one column, but will depend the search you want.
If you just only want to know the persons that is part of a group then you can store persons in one column with a group id as key. For update you just update the entire list in a group.
But if you want to search a specified person that belongs to group, then its not recommended that you store this multiple persons in one column. In this case its better to store a itermedium table that store person id, and group id.
Sounds like you want a table that maps users to group IDs and a second table that maps group IDs to which computers are in that group. I'm not sure, your language describing the problem was a bit confusing to me.
a list has some columns like: name, family name, phone number etc.
and rows like name=john familyName= lee number=12321321
name=... familyname=... number=...
an sql database works same way. every row in a sql database is a record. so you jusr add records of your list into your database using insert query.
complete explanation in here:
http://www.w3schools.com/sql/sql_insert.asp
This sounds like a typical many-to-many problem. You have many groups and many computers and they are related to eachother. In this situation, it is often recommended to use a mapping table, a.k.a. "junction table" or "cross-reference" table. This table consist solely of the two foreign keys in your other tables.
If your tables look like this:
Computer
- computerId
- otherComputerColumns
Group
- groupId
- othergroupColumns
Then your mapping table would look like this:
GroupComputer
- groupId
- computerId
And you would insert a single record for every relationship between a group and computer. This is in compliance with the rules for third normal form in regards to database normalization.
You can have a table with the group and group id, another table with the computer and computer id and a third table with the relation of group id and computer id.
I would like to store people's information in the database (table). Every person can have a different set of properties. I would like to store all these properties. But creating fixed number of columns does not make my application scalable.
So, another approach is to store these values inside key-value tables which leads to tables with a few columns but a huge number of rows.
So I am wondering if there is another way of storing these information which is also easy and fast to query.
What Database are you using.
One solution (if your DB supports) could be storing the person's information in an XML Format.
If you are stuck using SQL Server for this task, you can leverage its XML Support. Notably, you can use XPath.
Using XPath Queries in SQLXML 4.0
No matter how your data is organized conceptually, its the DBSM's job to make it fast, so usually, you shouldn't have to worry about whether your query is fast.
The first and most obvious, but also the least recommended way is using alter table which will allow you to modify the table later on.
The more recommended route is decomposing your table into many tables and relating them with primary keys.
So your person might have as a primary key social security, and attributes name, last name, street address, street name, postal code
Then you would decompose the table into (in pseudocode)
Table Person: social
Table Address: social, street_address, street_name, postal_code
Table Info: social, first_name, last_name
Then you can join them with
Select Person.social_security, first_name, last_name [...etc...] from Person, Address, Info where Person.social = Address.social and Person.social = Info.social
Then you can keep adding more tables as needed. There's no problem with having lots of rows in the other tables, and only a few in Person, and this is the recommended way because it lessens the number of empty spaces if a person doesn't have an address but does have a first name and last name, and for many other reasons.
My friend is building a product to be used by different independent medical units.
The database stores a vast collection of measurements taken at different times, like the temperature, blood pressure, etc...
Let us assume these are held in a table called exams with columns temperature, pressure, etc... (as well as id, patient_id and timestamp). Most of the measurements are stored as floats, but some are of other types (strings, integers...)
While many of these measurements are handled by their product, it needs to allow the different medical units to record and process other custom measurements. A very nifty UI allows the administrator to edit these customs fields, specify their name, type, possible range of values, etc...
He is unsure as to how to store these custom fields.
He is leaning towards a separate table (say a table custom_exam_data with fields like exam_id, custom_field_id, float_value, string_value, ...)
I worry that this will make searching both more difficult to achieve and less efficient.
I am leaning towards modifying the exam table directly (while avoiding conflicts on column names with some scheme like prefixing all custom fields with an underscore or naming them custom_1, ...)
He worries about modifying the database dynamically and having different schemas for each medical unit.
Hopefully some people which more experience can weigh in on this issue.
Notes:
he is using Ruby on Rails but I think this question is pretty much framework agnostic, except from the fact that he is only looking for solutions in SQL databases only.
I simplified the problem a bit since the custom fields need to be available for more than one table, but I believe this doesn`t really impact the direction to take.
(added) A very generic reporting module will need to search, sort, generate stats, etc.. of this data, so it is required that this data be stored in the columns of the appropriate type
(added) User inputs will be filtered, for the standard fields as well as for the custom fields. For example, numbers will be checked within a given range (can't have a temperature of -12 or +444), etc... Thus, conversion to the appropriate SQL type is not a problem.
I've had to deal with this situation many times over the years, and I agree with your initial idea of modifying the DB tables directly, and using dynamic SQL to generate statements.
Creating string UserAttribute or Key/Value columns sounds appealing at first, but it leads to the inner-platform effect where you end up having to re-implement foreign keys, data types, constraints, transactions, validation, sorting, grouping, calculations, et al. inside your RDBMS. You may as well just use flat files and not SQL at all.
SQL Server provides INFORMATION_SCHEMA tables that let you create, query, and modify table schemas at runtime. This has full type checking, constraints, transactions, calculations, and everything you need already built-in, don't reinvent it.
It's strange that so many people come up with ad-hoc solutions for this when there's a well-documented pattern for it:
Entity-Attribute-Value (EAV) Model
Two alternatives are XML and Nested Sets. XML is easier to manage but generally slow. Nested Sets usually require some type of proprietary database extension to do without making a mess, like CLR types in SQL Server 2005+. They violate first-normal form, but are nevertheless the fastest-performing solution.
Microsoft Dynamics CRM achieves this by altering the database design each time a change is made. Nasty, I think.
I would say a better option would be to consider an attribute table. Even though these are often frowned upon, it gives you the flexibility you need, and you can always create views using dynamic SQL to pivot the data out again. Just make sure you always use LEFT JOINs and FKs when creating these views, so that the Query Optimizer can do its job better.
I have seen a use of your friend's idea in a commercial accounting package. The table was split into two, first contained fields solely defined by the system, second contained fields like USER_STRING1, USER_STRING2, USER_FLOAT1 etc. The tables were linked by identity value (when a record is inserted into the main table, a record with same identity is inserted into the second one). Each table that needed user fields was split like that.
Well, whenever I need to store some unknown type in a database field, I usually store it as String, serializing it as needed, and also store the type of the data.
This way, you can have any kind of data, working with any type of database.
I would be inclined to store the measurement in the database as a string (varchar) with another column identifying the measurement type. My reasoning is that it will presumably, come from the UI as a string and casting to any other datatype may introduce a corruption before the user input get's stored.
The downside is that when you go to filter result-sets by some measurement metric you will still have to perform a casting but at least the storage and persistence mechanism is not introducing corruption.
I can't tell you the best way but I can tell you how Drupal achieves a sort of schemaless structure while still using the standard RDBMSs available today.
The general idea is that there's a schema table with a list of fields. Each row really only has two columns, the 'table':String column and the 'column':String column. For each of these columns it actually defines a whole table with just an id and the actual data for that column.
The trick really is that when you are working with the data it's never more than one join away from the bundle table that lists all the possible columns so you end up not losing as much speed as you might otherwise think. This will also allow you to expand much farther than just a few medical companies unlike the custom_ prefix you were proposing.
MySQL is very fast at returning row data for short rows with few columns. In this way this scheme ends up fairly quick while allowing you lots of flexibility.
As to search, my suggestion would be to index the page content instead of the database content. Use Solr to parse through rendered pages and hold links to the actual page instead of trying to search through the database using clever SQL.
Define two new tables: custom_exam_schema and custom_exam_data.
custom_exam_data has an exam_id column, plus an additional column for every custom attribute.
custom_exam_schema would have a row to describe how to interpret each of the columns of the custom_exam_data table. It would have columns like name, type, minValue, maxValue, etc.
So, for example, to create a custom field to track the number of fingers a person has, you would add ('fingerCount', 'number', 0, 10) to custom_exam_schema and then add a column named fingerCount to the exam table.
Someone might say it's bad to change the database schema at run time, but I'd argue that configuring these custom fields is part of set up and won't happen too often. Still, this method lets you handle changes at any time and doesn't risk messing around with your core table schemas.
lets say that your friend's database has to store data values from multiple sources such as demogrphic values, diagnosis, interventions, physionomic values, physiologic exam values, hospitalisation values etc.
He might have as well to define choices, lets say his database is missing the race and the unit staff need the race of the patient (different races are more unlikely to get some diseases), they might want to use a drop down with several choices.
I would propose to use an other table that would have these choices or would you just use a "Custom_field_choices" table, which at some point is exactly the same but with a different name.
Considering that the database :
- needs to be flexible
- that data from multiple tables can be added and be customized
- that you might want to keep the integrity of the main structure of your database for distribution and uniformity purpose
- that data MUST have a limit and alarms and warnings
- that data must have units ( 10 kg or 10 pounds) ?
- that data can have a selection of choices
- that data can be with different rights (from simple user to admin)
- that these data might be needed to generate reports without modifying the code (automation)
- that these data might be needed to make cross reference analysis within the system without modifying the code
the custom table would be my solution, modifying each table would end up being too risky.
I would store those custom fields in a table where each record ( dataType, dataValue, dataUnit ) would use in one row. So there would be a relation oneToMany from one sample to the data. You can also create a table to record all the kind of cutsom types you would use. For example:
create table DataType
(
id int primary key,
name varchar(100) not null unique
description text,
uri varchar(255) //<-- can be used for an ONTOLOGY
)
create table DataRecord
(
id int primary key,
sample_id int not null,//<-- reference to the sample
dataType_id int not null, //<-- references DataType
value varchar(100),//<-- the value as string
unit varchar(50)//<-- g, mg/ml, etc... but it could also be a link to a table describing the units just like DataType
)