Question

The below table has PC asset information and I need to remove slices of data from it based on different criteria.

I need to create a View in SQL Server 2005 which returns my results.

I tried to accomplish my goals using temporary tables until I realized that I could not use temporary tables in a View.

I then tried to use a CTE until I realized that deleting data from a CTE would also delete data from the actual table.

I cannot delete data from the actual table. I cannot create another table in the database either.

The table has 160,000 records.

The table:

TABLE dsm_hardware_basic
(
[UUID] binary(16) -- Randomly generated 16 digit key that is unique for each record, only column with no duplicate rows.
[HostUUID] binary(16) -- Randomly generated 16 digit key, column has duplicate rows.
[Name] nvarchar(255) -- Column that contains hostnames of computer assets. Example of record: PCASSET001. Column has duplicate rows.
[LastAgentExecution] datetime -- The last time that the software agent that collects asset information ran on the PC.
[HostName] nvarchar(255) -- The fully qualified domain name of the PC. Example of record: PCASSET001.companydomain.com. Column has duplicate rows.
)

I will explain what I want to accomplish:

1) Read in all the information from the table dbo.dsm_hardware_basic. Lets call this: dsm_hardware_basic_copy.

2) Query dbo.dsm_hardware_basic and remove data that fits the following criteria from dsm_hardware_basic_copy.
This basically removes the duplicate [HostUUID] with the oldest [LastAgentExecution] time.:

    SELECT   ,dsm_hardware_basic.[HostUUID]
             ,MIN(dsm_hardware_basic.[LastAgentExecution]) AS [LastAgentExecution]
    FROM      dsm_hardware_basic
    WHERE     dsm_hardware_basic.[HostUUID] <> ''
    GROUP BY  dsm_hardware_basic.[HostUUID]
              HAVING COUNT(*) = 2 -- The tiny amount of rows where this count is >2 will be left alone.

3) Additionaly query dbo.dsm_hardware_basic and remove data that fits the following criteria from dsm_hardware_basic_copy:
This basically removes the duplicate [HostName] with the oldest [LastAgentExecution] time.:

    SELECT   ,dsm_hardware_basic.[HostName]
             ,MIN(dsm_hardware_basic.[LastAgentExecution]) AS [LastAgentExecution]
    FROM      dsm_hardware_basic
    WHERE     dsm_hardware_basic.[HostName] <> ''
    GROUP BY  dsm_hardware_basic.[HostName]
              HAVING COUNT(*) > 1

I wasn’t sure how to do this in the above select, but not only should the COUNT of [HostName] be > 1, but [Name] should equal everything in [HostName] before the first period in [HostName]. Example [Name]: PCASSET001. Example [HostName]: PCASSET001.companydomain.com. I know this sounds strange considering the kind of PC data we are talking about in these two columns, but it is something I actually need to contend with.

3) Additionally query dbo.dsm_hardware_basic and remove data that fits the following criteria from dsm_hardware_basic_copy:

This basically removes the duplicate [Name] with the oldest [LastAgentExecution] time.:

    SELECT   ,dsm_hardware_basic.[Name]
             ,MIN(dsm_hardware_basic.[LastAgentExecution]) AS [LastAgentExecution]
    FROM      dsm_hardware_basic
    WHERE     dsm_hardware_basic.[Name] <> ''
    GROUP BY  dsm_hardware_basic.[Name]
              HAVING COUNT(*) = 2 -- The tiny amount of rows where this count is >2 will be left alone.

Answer 1

You’ve actually asked several different questions here and I’m not sure I completely follow the logic of the query, however, constructing it should not be too difficult.

To start with, you can work dsm_hardware_basic directly rather than a copy:

SELECT 
    * 
FROM dsm_hardware_basic

Now the part that

removes the duplicate [HostUUID] with the oldest [LastAgentExecution]
time

SELECT 
    dsm_hardware_basic.* 
FROM dsm_hardware_basic
INNER JOIN 
    (
        SELECT [UUID], ROW_NUMBER() OVER 
            (PARTITION BY [HostUUID] 
             ORDER BY [LastAgentExecution] DESC) AS host_UUID_rank
        FROM dsm_hardware_basic
        WHERE 
            [HostUUID] <> ''
    ) AS 
    duplicate_host_UUID_filtered ON dsm_hardware_basic.UUID = duplicate_host_UUID_filtered.UUID
    AND duplicate_host_UUID_filtered.host_UUID_rank = 1

What we’ve done is partitioned your table by HostUUID sorted by newest LastAgentExecution and removed every UUID from the query that matches our result using a JOIN.

We can now apply the same logic to your HostName:

SELECT 
    dsm_hardware_basic.* 
FROM dsm_hardware_basic
INNER JOIN 
    (
        SELECT [UUID], ROW_NUMBER() OVER 
            (PARTITION BY [HostUUID] 
             ORDER BY [LastAgentExecution] DESC) AS host_UUID_rank
        FROM dsm_hardware_basic
        WHERE 
            [HostUUID] <> ''
    ) AS 
    duplicate_host_UUID_filtered ON dsm_hardware_basic.UUID = duplicate_host_UUID_filtered.UUID
    AND duplicate_host_UUID_filtered.host_UUID_rank = 1
INNER JOIN 
    (
        SELECT [UUID], ROW_NUMBER() OVER 
            (PARTITION BY [HostName] 
             ORDER BY [LastAgentExecution] DESC) AS host_UUID_rank
        FROM dsm_hardware_basic
        WHERE 
            [HostName] <> ''
    ) AS 
    duplicate_HostName_filtered ON dsm_hardware_basic.UUID = duplicate_HostName_filtered.UUID
    AND duplicate_HostName_filtered.host_UUID_rank = 1

I’ll leave the final part to you as an exercise. Finally, after you’ve done debugging, just add CREATE VIEW to this.