For a new developement, I will have a big SQL table (~100M rows).
4 fields will be used to query the data.
Is it better to query one concatenated field with between or several equals ?
Exemple :
MainTable
PkId | Label | FkId1 | FkId2 | FkId3 | FkId4
1 | test | 1 | 4 | 3 | 1
Datas in Fk tables are static, example :
FkTable1
Id | Value
1 | a
2 | b
3 | c
To query the datas, the classic sql query is :
select Label, FkId1, FkId2, FkId3, FkId4
from MainTable
where FkId1=1 and FkId2=2 and FkId3 in(2, 3)
The idea to optimize performance is to add one field "UniqueId" calculated backend before the insert :
UniqueId = FkId1*1000000 + FkId2*10000 + FkId3*100 + FkId4
PkId | Label | FkId1 | FkId2 | FkId3 | FkId4 | UniqueId
1 | test | 1 | 4 | 3 | 1 | 1040301
select Label, FkId1, FkId2, FkId3, FkId4
from MainTable
where UniqueId between 1020200 and 1040000
Moreover, with the UniqueId field, an index on this field only will be sufficient.
What do you think ?
Thanks
For this query:
select Label, FkId1, FkId2, FkId3, FkId4
from MainTable
where FkId1 = 1 and FkId2 = 2 and FkId3 in (2, 3)
The optimal index is on MainTable(FkID1, FkId2, FkId3). You can also add Label and FkId4 to the index if you want a covering index (so the index can handle the entire query without referring to the original data pages).
There is no need for a computed field for the example you provided.
Since you will have 100M rows, thinking about optimisations from the start seems sensible to me.
However, your proposed solution will not work in this way:
Your formula above has two times the SAME factor 10000. You have to use different factors, i.e. different powers of 10.
Your select example has a "IN" clause (FkId3 in(2, 3)). This will only work, if only one of the FKs is queried this way. This fk should be the one with no factor in the formula for computing UniqueId (i.e. gives the least significant Digits of UniqueId).
Now seeing Gordons answer, I agree with him, i.e. using a combined index may be good enough for you (though your solution would probably slightly better). However, also the combined index has a similar problem: The FK field beeing queried with the IN clause should be the last field in the index.
I have a static dataset that correlates a range of numbers to some metadata, e.g.
+--------+--------+-------+--------+----------------+
| Min | Max |Country|CardType| Issuing Bank |
+--------+--------+-------+--------+----------------+
| 400011 | 400051 | USA |VISA | Bank of America|
+--------+--------+-------+--------+----------------+
| 400052 | 400062 | UK |MAESTRO | HSBC |
+--------+--------+-------+--------+----------------+
I wish to lookup a the data for some arbitrary single value
SELECT *
FROM SomeTable
WHERE Min <= 400030
AND Max >= 400030
I have about 200k of these range mappings, and am wondering the best table structure for SQL Server?
A composite key doesn't seem correct due to the fact that most of the time, the value being looked up will be in between the two range values stored on disk. Similarly, only indexing the first column doesn't seem to be selective enough.
I know that 200k rows is fairly insignificant, and I can get by with doing not much, but lets assume that the numbers of rows could be orders of magnitude greater.
If you usually search on both min and max then a compound key on (min,max) is appropriate. The engine will find all rows where min is less than X, then search within those result to find the rows where max is greater then Y.
The index would also be useful if you do searches on min only, but would not be applicable if you do searches only on max.
You can index the first number and then do the lookup like this:
select t.*,
(select top 1 s.country
from static s
where t.num >= s.firstnum
order by s.firstnum
) country
from sometable t;
Or use outer apply:
select t.*, s.country
from sometable t outer apply
(select top 1 s.country
from static s
where t.num >= s.firstnum
order by s.firstnum
) s
This should take advantage of an index on static(firstnum) or static(firstnum, country). This does not check against the second number. If that is important, use outer apply and do the check outside the subquery.
I would specify the primary key on (Min,Max). Queries are as simple as:
SELECT *
FROM SomeTable
WHERE #Value BETWEEN Min AND Max
I'd also define a constraint to enforce that Min <= Max. Then I would create a trigger to enforce uniqueness in ranges and prevent the database from storing an overlapping range.
I belive is easy/faster if you create a trigger for INSERT and then fill the related calculated columns country, issuing bank, card-number length
At the end you do the calculation only once, instead 200k every time you will do a query. Of course is there a space cost. But query will be much easier to mantain.
I remember once I have to calculate some sin and cos to calculate distance so I just create the calculated columns once.
After your update I think is even easier
+--------+--------+-------+--------+----------------+----------+
| Min | Max |Country|CardType| Issuing Bank | TypeID |
+--------+--------+-------+--------+----------------+----------+
| 400011 | 400051 | USA |VISA | Bank of America| 1 |
+--------+--------+-------+--------+----------------+----------+
| 400052 | 400062 | UK |MAESTRO | HSBC | 2 |
+--------+--------+-------+--------+----------------+----------+
Then you Card will also create a column TypeID
I have a table with the following structure
|user_id | place | type_of_place | money_earned| time |
|--------+-------+---------------+-------------+------|
| | | | | |
The table is very large, several millions of rows. The data is in a PostgreSQL 9.1 database.
I want to calculate, per user_id and type_of_place: the mean, the standard deviation, and the top 5 of places (ordered by counts), and the most used hour of time (mode).
The resulting data must be in this form:
| user_id | type_of_place | avg | stddev | top5_places | mode |
+---------+---------------+-----+--------+------------------+------+
| 1 | tp1 | 10 | 1 | {p1,p2,p3,p4,p5} | 8 |
| 2 | tp1 | 3 | 2 | {p3,p4} | 23 |
| 1 | tp3 | 1 | 1 | {p1} | 4 |
etc.
Is there a for of doing this with window functions efficiently?
What if I want to grouping by week? (i.e. another column that represents the number of week)
Thank you!
A standard GROUP BY query will get you most of the way:
SELECT
user_id,
type_of_place,
avg(money_earned) AS avg,
stddev(money_earned) AS stddev
FROM
earnings -- I'm not sure what your data table is called...
GROUP BY
user_id,
type_of_place
This leaves the top5_places and mode columns. These are both also aggregates, but not ones which are defined in the standard PostgreSQL installation. Luckily, you can add them.
Here's a page discussing how to define a mode aggregate function: http://wiki.postgresql.org/wiki/Aggregate_Mode
Once you have a mode aggregate function, assuming time is a timestamp of some kind, the expression you will add to the select list will be:
SELECT
...
mode(extract(hour FROM time)) AS mode -- Add this expression
FROM
...
Assuming order by money
For top5_places, there are several approaches, but the quickest is probably to use PostgreSQL's builtin array_agg function, and take the first 5 elements:
SELECT
...
(array_agg(place ORDER BY money_earned DESC))[1:5] AS top5_places -- Add this expression
FROM
...
One alternative is to define another aggregate called (for instance) top5, which performs the same function. This could be more efficient if there are many distinct places for each user/type of place combination, since it can stop accumulating after the first 5, whereas the above expression will generally build a complete array of all places, and then truncate to the first 5.
This assumes that a place has a unique earnings entry for each user/type combination. If a place can occur more than once, and you want to sort by sum(money_earned) for each place, then you need to use a subquery like in the examples below...
Order by counts
Ok, so the places should be ordered by how often they occur. Here's a quick way, which uses a couple of subqueries -- add this as an expression to the select-clause of the above query:
(SELECT
(array_agg(place ORDER BY cnt DESC))[1:5]
FROM
(SELECT place, count(*) FROM earnings AS t2
WHERE t2.user_id = earnings.user_id AND t2.type_of_place = earnings.type_of_place
GROUP BY place) AS s (place, cnt)
) AS top5_places
The inner subquery called s evaluates to a table of each place for that user/type combination, and the number of times it occurs (which I've called cnt). These are then fed to array_agg in descending order of that count.
I suspect there could be much neater (and probably more efficient) ways of writing it. If not, then I would recommend trying to move this complicated expression into a function or aggregate, if you can...
Histrogram of places in each hour
We'll use a similar expression, which will return the array of counts, ordered by hour:
(SELECT
array_agg(cnt ORDER BY hour DESC)
FROM
(SELECT extract(hour FROM time), count(*) FROM earnings AS t2
WHERE t2.user_id = earnings.user_id AND t2.type_of_place = earnings.type_of_place
GROUP BY 1) AS s (hour, cnt)
) AS hourly_histogram
(Add that to the select-clause of the original query.)
I learned something simple about SQL the other day:
SELECT c FROM myTbl GROUP BY C
Has the same result as:
SELECT DISTINCT C FROM myTbl
What I am curious of, is there anything different in the way an SQL engine processes the command, or are they truly the same thing?
I personally prefer the distinct syntax, but I am sure it's more out of habit than anything else.
EDIT: This is not a question about aggregates. The use of GROUP BY with aggregate functions is understood.
MusiGenesis' response is functionally the correct one with regard to your question as stated; the SQL Server is smart enough to realize that if you are using "Group By" and not using any aggregate functions, then what you actually mean is "Distinct" - and therefore it generates an execution plan as if you'd simply used "Distinct."
However, I think it's important to note Hank's response as well - cavalier treatment of "Group By" and "Distinct" could lead to some pernicious gotchas down the line if you're not careful. It's not entirely correct to say that this is "not a question about aggregates" because you're asking about the functional difference between two SQL query keywords, one of which is meant to be used with aggregates and one of which is not.
A hammer can work to drive in a screw sometimes, but if you've got a screwdriver handy, why bother?
(for the purposes of this analogy, Hammer : Screwdriver :: GroupBy : Distinct and screw => get list of unique values in a table column)
GROUP BY lets you use aggregate functions, like AVG, MAX, MIN, SUM, and COUNT.
On the other hand DISTINCT just removes duplicates.
For example, if you have a bunch of purchase records, and you want to know how much was spent by each department, you might do something like:
SELECT department, SUM(amount) FROM purchases GROUP BY department
This will give you one row per department, containing the department name and the sum of all of the amount values in all rows for that department.
What's the difference from a mere duplicate removal functionality point of view
Apart from the fact that unlike DISTINCT, GROUP BY allows for aggregating data per group (which has been mentioned by many other answers), the most important difference in my opinion is the fact that the two operations "happen" at two very different steps in the logical order of operations that are executed in a SELECT statement.
Here are the most important operations:
FROM (including JOIN, APPLY, etc.)
WHERE
GROUP BY (can remove duplicates)
Aggregations
HAVING
Window functions
SELECT
DISTINCT (can remove duplicates)
UNION, INTERSECT, EXCEPT (can remove duplicates)
ORDER BY
OFFSET
LIMIT
As you can see, the logical order of each operation influences what can be done with it and how it influences subsequent operations. In particular, the fact that the GROUP BY operation "happens before" the SELECT operation (the projection) means that:
It doesn't depend on the projection (which can be an advantage)
It cannot use any values from the projection (which can be a disadvantage)
1. It doesn't depend on the projection
An example where not depending on the projection is useful is if you want to calculate window functions on distinct values:
SELECT rating, row_number() OVER (ORDER BY rating) AS rn
FROM film
GROUP BY rating
When run against the Sakila database, this yields:
rating rn
-----------
G 1
NC-17 2
PG 3
PG-13 4
R 5
The same couldn't be achieved with DISTINCT easily:
SELECT DISTINCT rating, row_number() OVER (ORDER BY rating) AS rn
FROM film
That query is "wrong" and yields something like:
rating rn
------------
G 1
G 2
G 3
...
G 178
NC-17 179
NC-17 180
...
This is not what we wanted. The DISTINCT operation "happens after" the projection, so we can no longer remove DISTINCT ratings because the window function was already calculated and projected. In order to use DISTINCT, we'd have to nest that part of the query:
SELECT rating, row_number() OVER (ORDER BY rating) AS rn
FROM (
SELECT DISTINCT rating FROM film
) f
Side-note: In this particular case, we could also use DENSE_RANK()
SELECT DISTINCT rating, dense_rank() OVER (ORDER BY rating) AS rn
FROM film
2. It cannot use any values from the projection
One of SQL's drawbacks is its verbosity at times. For the same reason as what we've seen before (namely the logical order of operations), we cannot "easily" group by something we're projecting.
This is invalid SQL:
SELECT first_name || ' ' || last_name AS name
FROM customer
GROUP BY name
This is valid (repeating the expression)
SELECT first_name || ' ' || last_name AS name
FROM customer
GROUP BY first_name || ' ' || last_name
This is valid, too (nesting the expression)
SELECT name
FROM (
SELECT first_name || ' ' || last_name AS name
FROM customer
) c
GROUP BY name
I've written about this topic more in depth in a blog post
There is no difference (in SQL Server, at least). Both queries use the same execution plan.
http://sqlmag.com/database-performance-tuning/distinct-vs-group
Maybe there is a difference, if there are sub-queries involved:
http://blog.sqlauthority.com/2007/03/29/sql-server-difference-between-distinct-and-group-by-distinct-vs-group-by/
There is no difference (Oracle-style):
http://asktom.oracle.com/pls/asktom/f?p=100:11:0::::P11_QUESTION_ID:32961403234212
Use DISTINCT if you just want to remove duplicates. Use GROUPY BY if you want to apply aggregate operators (MAX, SUM, GROUP_CONCAT, ..., or a HAVING clause).
I expect there is the possibility for subtle differences in their execution.
I checked the execution plans for two functionally equivalent queries along these lines in Oracle 10g:
core> select sta from zip group by sta;
---------------------------------------------------------------------------
| Id | Operation | Name | Rows | Bytes | Cost (%CPU)| Time |
---------------------------------------------------------------------------
| 0 | SELECT STATEMENT | | 58 | 174 | 44 (19)| 00:00:01 |
| 1 | HASH GROUP BY | | 58 | 174 | 44 (19)| 00:00:01 |
| 2 | TABLE ACCESS FULL| ZIP | 42303 | 123K| 38 (6)| 00:00:01 |
---------------------------------------------------------------------------
core> select distinct sta from zip;
---------------------------------------------------------------------------
| Id | Operation | Name | Rows | Bytes | Cost (%CPU)| Time |
---------------------------------------------------------------------------
| 0 | SELECT STATEMENT | | 58 | 174 | 44 (19)| 00:00:01 |
| 1 | HASH UNIQUE | | 58 | 174 | 44 (19)| 00:00:01 |
| 2 | TABLE ACCESS FULL| ZIP | 42303 | 123K| 38 (6)| 00:00:01 |
---------------------------------------------------------------------------
The middle operation is slightly different: "HASH GROUP BY" vs. "HASH UNIQUE", but the estimated costs etc. are identical. I then executed these with tracing on and the actual operation counts were the same for both (except that the second one didn't have to do any physical reads due to caching).
But I think that because the operation names are different, the execution would follow somewhat different code paths and that opens the possibility of more significant differences.
I think you should prefer the DISTINCT syntax for this purpose. It's not just habit, it more clearly indicates the purpose of the query.
For the query you posted, they are identical. But for other queries that may not be true.
For example, it's not the same as:
SELECT C FROM myTbl GROUP BY C, D
I read all the above comments but didn't see anyone pointed to the main difference between Group By and Distinct apart from the aggregation bit.
Distinct returns all the rows then de-duplicates them whereas Group By de-deduplicate the rows as they're read by the algorithm one by one.
This means they can produce different results!
For example, the below codes generate different results:
SELECT distinct ROW_NUMBER() OVER (ORDER BY Name), Name FROM NamesTable
SELECT ROW_NUMBER() OVER (ORDER BY Name), Name FROM NamesTable
GROUP BY Name
If there are 10 names in the table where 1 of which is a duplicate of another then the first query returns 10 rows whereas the second query returns 9 rows.
The reason is what I said above so they can behave differently!
If you use DISTINCT with multiple columns, the result set won't be grouped as it will with GROUP BY, and you can't use aggregate functions with DISTINCT.
GROUP BY has a very specific meaning that is distinct (heh) from the DISTINCT function.
GROUP BY causes the query results to be grouped using the chosen expression, aggregate functions can then be applied, and these will act on each group, rather than the entire resultset.
Here's an example that might help:
Given a table that looks like this:
name
------
barry
dave
bill
dave
dave
barry
john
This query:
SELECT name, count(*) AS count FROM table GROUP BY name;
Will produce output like this:
name count
-------------
barry 2
dave 3
bill 1
john 1
Which is obviously very different from using DISTINCT. If you want to group your results, use GROUP BY, if you just want a unique list of a specific column, use DISTINCT. This will give your database a chance to optimise the query for your needs.
If you are using a GROUP BY without any aggregate function then internally it will treated as DISTINCT, so in this case there is no difference between GROUP BY and DISTINCT.
But when you are provided with DISTINCT clause better to use it for finding your unique records because the objective of GROUP BY is to achieve aggregation.
They have different semantics, even if they happen to have equivalent results on your particular data.
Please don't use GROUP BY when you mean DISTINCT, even if they happen to work the same. I'm assuming you're trying to shave off milliseconds from queries, and I have to point out that developer time is orders of magnitude more expensive than computer time.
In Teradata perspective :
From a result set point of view, it does not matter if you use DISTINCT or GROUP BY in Teradata. The answer set will be the same.
From a performance point of view, it is not the same.
To understand what impacts performance, you need to know what happens on Teradata when executing a statement with DISTINCT or GROUP BY.
In the case of DISTINCT, the rows are redistributed immediately without any preaggregation taking place, while in the case of GROUP BY, in a first step a preaggregation is done and only then are the unique values redistributed across the AMPs.
Don’t think now that GROUP BY is always better from a performance point of view. When you have many different values, the preaggregation step of GROUP BY is not very efficient. Teradata has to sort the data to remove duplicates. In this case, it may be better to the redistribution first, i.e. use the DISTINCT statement. Only if there are many duplicate values, the GROUP BY statement is probably the better choice as only once the deduplication step takes place, after redistribution.
In short, DISTINCT vs. GROUP BY in Teradata means:
GROUP BY -> for many duplicates
DISTINCT -> no or a few duplicates only .
At times, when using DISTINCT, you run out of spool space on an AMP. The reason is that redistribution takes place immediately, and skewing could cause AMPs to run out of space.
If this happens, you have probably a better chance with GROUP BY, as duplicates are already removed in a first step, and less data is moved across the AMPs.
group by is used in aggregate operations -- like when you want to get a count of Bs broken down by column C
select C, count(B) from myTbl group by C
distinct is what it sounds like -- you get unique rows.
In sql server 2005, it looks like the query optimizer is able to optimize away the difference in the simplistic examples I ran. Dunno if you can count on that in all situations, though.
In that particular query there is no difference. But, of course, if you add any aggregate columns then you'll have to use group by.
You're only noticing that because you are selecting a single column.
Try selecting two fields and see what happens.
Group By is intended to be used like this:
SELECT name, SUM(transaction) FROM myTbl GROUP BY name
Which would show the sum of all transactions for each person.
From a 'SQL the language' perspective the two constructs are equivalent and which one you choose is one of those 'lifestyle' choices we all have to make. I think there is a good case for DISTINCT being more explicit (and therefore is more considerate to the person who will inherit your code etc) but that doesn't mean the GROUP BY construct is an invalid choice.
I think this 'GROUP BY is for aggregates' is the wrong emphasis. Folk should be aware that the set function (MAX, MIN, COUNT, etc) can be omitted so that they can understand the coder's intent when it is.
The ideal optimizer will recognize equivalent SQL constructs and will always pick the ideal plan accordingly. For your real life SQL engine of choice, you must test :)
PS note the position of the DISTINCT keyword in the select clause may produce different results e.g. contrast:
SELECT COUNT(DISTINCT C) FROM myTbl;
SELECT DISTINCT COUNT(C) FROM myTbl;
I know it's an old post. But it happens that I had a query that used group by just to return distinct values when using that query in toad and oracle reports everything worked fine, I mean a good response time. When we migrated from Oracle 9i to 11g the response time in Toad was excellent but in the reporte it took about 35 minutes to finish the report when using previous version it took about 5 minutes.
The solution was to change the group by and use DISTINCT and now the report runs in about 30 secs.
I hope this is useful for someone with the same situation.
Sometimes they may give you the same results but they are meant to be used in different sense/case. The main difference is in syntax.
Minutely notice the example below. DISTINCT is used to filter out the duplicate set of values. (6, cs, 9.1) and (1, cs, 5.5) are two different sets. So DISTINCT is going to display both the rows while GROUP BY Branch is going to display only one set.
SELECT * FROM student;
+------+--------+------+
| Id | Branch | CGPA |
+------+--------+------+
| 3 | civil | 7.2 |
| 2 | mech | 6.3 |
| 6 | cs | 9.1 |
| 4 | eee | 8.2 |
| 1 | cs | 5.5 |
+------+--------+------+
5 rows in set (0.001 sec)
SELECT DISTINCT * FROM student;
+------+--------+------+
| Id | Branch | CGPA |
+------+--------+------+
| 3 | civil | 7.2 |
| 2 | mech | 6.3 |
| 6 | cs | 9.1 |
| 4 | eee | 8.2 |
| 1 | cs | 5.5 |
+------+--------+------+
5 rows in set (0.001 sec)
SELECT * FROM student GROUP BY Branch;
+------+--------+------+
| Id | Branch | CGPA |
+------+--------+------+
| 3 | civil | 7.2 |
| 6 | cs | 9.1 |
| 4 | eee | 8.2 |
| 2 | mech | 6.3 |
+------+--------+------+
4 rows in set (0.001 sec)
Sometimes the results that can be achieved by GROUP BY clause is not possible to achieved by DISTINCT without using some extra clause or conditions. E.g in above case.
To get the same result as DISTINCT you have to pass all the column names in GROUP BY clause like below. So see the syntactical difference. You must have knowledge about all the column names to use GROUP BY clause in that case.
SELECT * FROM student GROUP BY Id, Branch, CGPA;
+------+--------+------+
| Id | Branch | CGPA |
+------+--------+------+
| 1 | cs | 5.5 |
| 2 | mech | 6.3 |
| 3 | civil | 7.2 |
| 4 | eee | 8.2 |
| 6 | cs | 9.1 |
+------+--------+------+
Also I have noticed GROUP BY displays the results in ascending order by default which DISTINCT does not. But I am not sure about this. It may be differ vendor wise.
Source : https://dbjpanda.me/dbms/languages/sql/sql-syntax-with-examples#group-by
In terms of usage, GROUP BY is used for grouping those rows you want to calculate. DISTINCT will not do any calculation. It will show no duplicate rows.
I always used DISTINCT if I want to present data without duplicates.
If I want to do calculations like summing up the total quantity of mangoes, I will use GROUP BY
In Hive (HQL), GROUP BY can be way faster than DISTINCT, because the former does not require comparing all fields in the table.
See: https://sqlperformance.com/2017/01/t-sql-queries/surprises-assumptions-group-by-distinct.
The way I always understood it is that using distinct is the same as grouping by every field you selected in the order you selected them.
i.e:
select distinct a, b, c from table;
is the same as:
select a, b, c from table group by a, b, c
Funtional efficiency is totally different.
If you would like to select only "return value" except duplicate one, use distinct is better than group by. Because "group by" include ( sorting + removing ) , "distinct" include ( removing )
Generally we can use DISTINCT for eliminate the duplicates on Specific Column in the table.
In Case of 'GROUP BY' we can Apply the Aggregation Functions like
AVG, MAX, MIN, SUM, and COUNT on Specific column and fetch
the column name and it aggregation function result on the same column.
Example :
select specialColumn,sum(specialColumn) from yourTableName group by specialColumn;
There is no significantly difference between group by and distinct clause except the usage of aggregate functions.
Both can be used to distinguish the values but if in performance point of view group by is better.
When distinct keyword is used , internally it used sort operation which can be view in execution plan.
Try simple example
Declare #tmpresult table
(
Id tinyint
)
Insert into #tmpresult
Select 5
Union all
Select 2
Union all
Select 3
Union all
Select 4
Select distinct
Id
From #tmpresult