Expand query result in select where statement - mysql

Question

I need to get all the values between two dates on the same table, but also, include the value immediately before the earliest one:
To be more clear, I have the following table (Table 1. with aprox. 46 million rows):
Table 1. (Table I have)
updatetime | value
. .
. .
2018-01-01 08:32:02 | 2
2018-01-01 09:23:12 | 5
2018-01-01 10:45:00 | 8 * value to include on the result.
2018-01-01 11:10:44 | 10 * <-- earliest date.
2018-01-01 13:11:54 | 12 *
2018-01-01 16:14:57 | 16 *
2018-01-01 20:00:55 | 22 *
2018-01-01 22:34:43 | 23 *
2018-01-01 23:55:23 | 23 * <-- latest date.
2018-01-02 01:03:57 | 25
2018-01-02 03:39:07 | 28
. .
. .
I have accumulative values, so, the next value will always be equal or greater than the one before. I need to get all values between 2018-01-01 11:10:44 and 2018-01-02 01:03:57, but, I also need the value before the earliest date, that means that I need this result: (Table 2.)
Table 2. (Result I need)
2018-01-01 10:45:00 | 8
2018-01-01 11:10:44 | 10
2018-01-01 13:11:54 | 12
2018-01-01 16:14:57 | 16
2018-01-01 20:00:55 | 22
2018-01-01 22:34:43 | 23
2018-01-01 23:55:23 | 23
As you could see in Table 1, date spaces are not the same, so, previous date of the earliest date is unknown to me, and value incrementation is also random.
I already tried theese two queries, but I have doubts about performance:
The first one is just two queries made one, very bad performance because of the change of order of the table.
SELECT * FROM his
WHERE updatetime>=(
SELECT updatetime
FROM definition
WHERE updatetime<"2018-01-01 11:10:44"
ORDER BY updatetime DESC
LIMIT 1
) and updatetime<="2018-01-02 01:03:57";
The second should be slower than the first one, but if I define the column as incremental to make MySQL walk the table from the last value of the condition and not to walk all over it?
SELECT * FROM his
WHERE updatetime>=(
SELECT MAX(updatetime)
FROM definition
WHERE updatetime<"2018-01-01 11:10:44"
LIMIT 1
) AND updatetime<="2018-01-02 01:03:57";

Answer 1

First of all, for these queries to perform efficiently, you would need an index on column updatetime :
CREATE INDEX idx_updatetime ON mytable(updatetime);
One way to do it would to use a correlated subquery in the WHERE clause that selects the previous record :
SELECT t.*
FROM mytable t
WHERE
t.updatetime < '2018-01-02 01:03:57'
AND NOT EXISTS (
SELECT 1
FROM mytable t1
WHERE t1.updatetime < '2018-01-01 11:10:44' AND t1.updatetime > t.updatetime
)
Demo on DB Fiddle :
| updatetime | value |
| ------------------- | ----- |
| 2018-01-01 11:10:44 | 10 |
| 2018-01-01 13:11:54 | 12 |
| 2018-01-01 16:14:57 | 16 |
| 2018-01-01 20:00:55 | 22 |
| 2018-01-01 22:34:43 | 23 |
| 2018-01-01 23:55:23 | 23 |
Another option is to compute the updatetime of the previous record in a subquery, and then use it to filter the table. This might perform better than the first query.
SELECT t.*
FROM mytable t
INNER JOIN (
SELECT MAX(updatetime) updatetime
FROM mytable
WHERE updatetime < '2018-01-01 11:10:44'
) x ON t.updatetime >= x.updatetime AND t.updatetime < '2018-01-02 01:03:57';
Demo on DB Fiddle