Teradata Forums - Database

Hi Guys,
I see that my database(ver:15.00.04.03) allows table names > 30 (I think 128 char limit is in place in our envronment) . But i still see the ET,LT,WT tables created during MLOAD still confine to 30 character limit . Is there a separate limit for these utility tables created at runtime ?.

I am not able to find any specific limits in the documentation. Can somebody let me know if there is a limit on these utility tables ?

HI,
Need help to with SQL to solve below.

My table contains tow columns Member number and elig date (which is always beginning of the month)
Member #    Elig_dt                           Expected Result (no. of months continuously eligible)
=============                         =============
M1             2015/01/01                         1
M1             2015/02/01                         2
M1             2015/03/01                         3
M1            2015/12/01                          1
M2            2015/01/01                          1
M2            2015/02/01                          2
M2           2015/12/01                           1
I need to count no. of months a member is eligibile without interruption. if interruption the counter should be reset to 1

Any help from any one is greatly appreciated.

Thx

how to get dense rank in teradata?
is there any dense_rank() function available in teradata

What's your Teradata release?
Since 14.10 there's LAST_VALUE:

select Member, Elig_dt,
                     -- find the previous gap's row number
   rn - last_value(case when flag = 1 then rn -1 end ignore nulls)
        over (partition by Member
              order by Elig_dt
              rows unbounded preceding)
from
 (
   select Member, Elig_dt,
      row_number()                -- row number without reset
      over (partition by Member
            order by Elig_dt) as rn,
      case when min(Elig_dt)      -- check for gap
                over (partition by Member
                      order by Elig_dt
                      rows between 1 preceding and 1 preceding)
                >= add_months(Elig_dt, -1)
           then 0                 -- no gap
           else 1                 -- gap
      end as flag
   from tab
 ) as dt

It's available since TD14.10, before you need to rewrite it:
Missing Functions: DENSE_RANK

There is no separate limit for load utility table names. How are you determining that there is a limitation?

Multi/Fast-Load are obsolete now. Can you try with TPT which is a replacement for them?

When you copy SQL from other tool (mostly) it can have some special characters. You might want to try pasting the SQL in Notepad++ or other tool which can show hidden characters as well.

SELECT CURRENT_DATE, CURRENT_DATE - 1, CURRENT_DATE - 2

The following should help resolving this issue:

https://forums.teradata.com/forum/database/using-merge-into-a-table-with-an-identity-column-result-failure-5758

LOCKING ROW FOR ACCESS will give you the dirty read. Check the Teradata documentation, there is a matrix of which locks can supercede and stuff. It would help.

How are you processing the BTEQ file? Doesn't seem to be a BTEQ issue rather perhaps the shell issue perhaps.

Looks good. Are you sure you want to continue after getting an exception?

In one line, there is no alternative to LEFT OUTER JOIN.
Regarding fixing performance .... How are the tables designed? What the joining columns? What is the PI? .... search forum and you will get many scenarios of performance issues.
There is no magic to performant queries. :)

Have you contacted Teradata CS? If no, you should. If yes, they must have an answer to why it happened.

Does your query include UNION/UNION ALL? Can you share your query?

Whats the structure of the table? Parent & child are in separate columns or in one? Have you tried ORDER BY Parent, Child?

On SQL Server you can do something like DateAdd(ms, DateDiff(ms, date1, date2), date1)/2
To get the mid date between two date 2016-02-03 12:00:00 and 2016-02-03 14:00:00, you should get the result 2016-02-03 13:00:00.
How do you do that in Teradata when you have two timestamp(6) values?
Peter Schwennesen

Tags:

Forums:

Dear all,
I have the following data:

DATE	     CODE	RANK
?	        ABS	0
12/04/2014	RET	0
20/04/2014	STT     0
01/05/2014	RETk	0
13/05/2014	RETj	0

RANK is the column I want to calculate in my SQL given the columns DATE and CODE. It's initialized here to 0.
The logic i want to implement is as follows:
If date = '?' (that is the first line)
If CODE = 'ABS' THEN 0 ELSE 1
Else
If RANK - 1 (previous RANK value) = 1 AND CODE = RET THEN RANK = 2
Else 0
I cannot figure out how to update the current column value given the previous one... I tried many logic implementation with OLAP functions without success.
Can anyone give me a hint?
Regards.

Tags:

need update logic

Forums:

Database

Hi Peter,
as long as the difference between the timestamps is less than 10000 days:

SELECT TIMESTAMP '2016-02-03 12:00:00' AS ts1, TIMESTAMP '2016-02-03 14:00:00' AS ts2,
   ts1 + ((ts2-ts1 DAY(4) TO SECOND)/2)

Latest Images