Tom.Thomson (11/3/2009)

---

Your integer representation flips the top bit, so it isn't a natural representation, and I already had posted a suggested mod to your transformation which delivers an integer without...

Tom.Thomson (10/31/2009)

---

It seemed to me to be probable that you were a DBA - after all, most of the articles posted on SQL ServerCentral are written by DBAs, and you...

Tom.Thomson (10/29/2009)

---

... he is a DBA not a data communications developer, but in my view DBAs should ask people with application field expertise before taking decisions like the choice of...

My intention was not to mislead or misinform, however do keep in mind that the article was dedicated to SQL Server efficiency. I am fully versed in how 32-bit network...

Also, the choice between varbinary(4) (which should really be binary(4)) and int comes down to deciding between being able to natively bitmask IPs with subnet from within SQL Server or...

Dynamically computed : (100000 row(s) affected)

SQL Server Execution Times:

CPU time = 204 ms, elapsed time = 4500 ms.

Persisted : (100000 row(s) affected)

SQL Server Execution Times:

...

Elliott W (8/31/2009)

---

Also, it looks like Jean-Sebastien's solution stores the text and calculates the int, in that case you spend up to 11 extra bytes to store the address. ...

I respectfully disagree. As has already been discussed, the performance hit is due to the use of UDFs. Creating a computed column on the parsename method demonstrates how fast parsename...

[ See after next post ... ]

I have no problem releasing this code under the LGPL. I'll make a mention as such when I made my addendum.

I'll post an addendum to the article soon discussing the combination of an integer column with a computed human readable column. In the meantime, I retested the four different methods...

When selecting data, I believe that combining my approach with Jeff's suggestions results in excellent performance. If the primary stored field for the IP address is still an int, we...

I must appologize Jeff, in my haste I misread part of your code. It is indeed an interesting approach that I will explore further.

I've seen a lot of posts regarding the poor performance of UDFs. I believe that's missing the point. Generally speaking, there is no need for displaying 10,000 IPv4 addresses in...

I believe that another distinction that must be taken into account is the difference between working with individual IP addresses and IP networks. Design considerations are vastly different for each...