Use DATENAME and wrap the logic in a Function, not a Stored Proc

declare @myTime as DateTime

set @myTime = GETDATE()

select @myTime

select DATENAME(day, @myTime) + SUBSTRING(UPPER(DATENAME(month, @myTime)), 0,4)

Returns "14OCT"

Try not to use any Character / String based operations if possible when working with dates. They are numerical (a float) and performance will suffer from those data type conversions.

Dig these handy conversions I have compiled over the years...

/* Common date functions */
--//This contains common date functions for MSSQL server

/*Getting Parts of a DateTime*/
    --//gets the date only, 20x faster than using Convert/Cast to varchar
    --//this has been especially useful for JOINS
    SELECT (CAST(FLOOR(CAST(GETDATE() as FLOAT)) AS DateTime))

    --//gets the time only (date portion is '1900-01-01' and is considered the "0 time" of dates in MSSQL, even with the datatype min value of 01/01/1753. 
    SELECT (GETDATE() - (CAST(FLOOR(CAST(GETDATE() as FLOAT)) AS DateTime)))


/*Relative Dates*/
--//These are all functions that will calculate a date relative to the current date and time
    /*Current Day*/
    --//now
    SELECT (GETDATE())

    --//midnight of today
    SELECT (DATEADD(ms,-4,(DATEADD(dd,DATEDIFF(dd,0,GETDATE()) + 1,0))))

    --//Current Hour
    SELECT DATEADD(hh,DATEPART(hh,GETDATE()),CAST(FLOOR(CAST(GETDATE() AS FLOAT)) as DateTime))

    --//Current Half-Hour - if its 9:36, this will show 9:30
    SELECT DATEADD(mi,((DATEDIFF(mi,(CAST(FLOOR(CAST(GETDATE() as FLOAT)) as DateTime)), GETDATE())) / 30) * 30,(CAST(FLOOR(CAST(GETDATE() as FLOAT)) as DateTime)))

    /*Yearly*/
    --//first datetime of the current year
    SELECT (DATEADD(yy,DATEDIFF(yy,0,GETDATE()),0))

    --//last datetime of the current year
    SELECT (DATEADD(ms,-4,(DATEADD(yy,DATEDIFF(yy,0,GETDATE()) + 1,0))))

    /*Monthly*/
    --//first datetime of current month
    SELECT (DATEADD(mm,DATEDIFF(mm,0,GETDATE()),0))

    --//last datetime of the current month
    SELECT (DATEADD(ms,-4,DATEADD(mm,1,DATEADD(mm,DATEDIFF(mm,0,GETDATE()),0))))

    --//first datetime of the previous month
    SELECT (DATEADD(mm,DATEDIFF(mm,0,GETDATE()) -1,0))

    --//last datetime of the previous month
    SELECT (DATEADD(ms, -4,DATEADD(mm,DATEDIFF(mm,0,GETDATE()),0)))

    /*Weekly*/
    --//previous monday at 12AM
    SELECT (DATEADD(wk,DATEDIFF(wk,0,GETDATE()) -1 ,0))

    --//previous friday at 11:59:59 PM
    SELECT (DATEADD(ms,-4,DATEADD(dd,5,DATEADD(wk,DATEDIFF(wk,0,GETDATE()) -1 ,0))))

    /*Quarterly*/
    --//first datetime of current quarter
    SELECT (DATEADD(qq,DATEDIFF(qq,0,GETDATE()),0))

    --//last datetime of current quarter
    SELECT (DATEADD(ms,-4,DATEADD(qq,DATEDIFF(qq,0,GETDATE()) + 1,0)))

You can use the following command in SQL server to make it:

select FORMAT(getdate(), N'yyyy-MM-ddThh:mm:ss')

If dt is your datetime column, then

For 1:

SUBSTRING(CONVERT(varchar, dt, 13), 1, 2)
    + UPPER(SUBSTRING(CONVERT(varchar, dt, 13), 4, 3))

For 2:

SUBSTRING(CONVERT(varchar, dt, 100), 13, 2)
    + SUBSTRING(CONVERT(varchar, dt, 100), 16, 3)

Not answering your question specifically, but isn't that something that should be handled by the presentation layer of your application. Doing it the way you describe creates extra processing on the database end as well as adding extra network traffic (assuming the database exists on a different machine than the application), for something that could be easily computed on the application side, with more rich date processing libraries, as well as being more language agnostic, especially in the case of your first example which contains the abbreviated month name. Anyway the answers others give you should point you in the right direction if you still decide to go this route.