I have inherited a table with a structure something like this:
ID Name Timestamp Data
----------------------------
1 A 40 ...
2 A
If you are using SQL Server 2005/2008, then the CTE solution already listed by Mitch Weat is the best from a performance perspective. However, if you are using SQL Server 2000, then you can't assume there aren't duplicate Name | TimeStamp combinations. Use the following code to return only one record per name:
SELECT ID
, Name
, TimeStamp
, Data
FROM DataTable dt
INNER JOIN
(SELECT Name
, MIN(DataTable.ID) AS MinimumID
FROM DataTable
INNER JOIN
(SELECT Name
, MAX(Timestamp) AS Timestamp
FROM DataTable
GROUP BY Name) latest
ON DataTable.Name = Latest.Name
AND DataTable.Timestamp = Latest.Timestamp
GROUP BY Name) MinimumLatest
ON dt.ID = MinimumLatest.ID
So if you add another record like 9 C 30, then this will only return ID 6. If you don't go this far, then you may end up return 9 C 30 and 6 C 30.