ScottPletcher (9/2/2015)

---

Lynn Pettis (9/2/2015)

---

ScottPletcher (9/2/2015)

---

Finally, there's the cost to fix each issue.

Once people are properly convinced that NOLOCK is a bad idea in many, many cases, they simply remove those specific bytes -- "WITH (NOLOCK)" -- from the code. That's the sum total of rework effort.

But for a bad identity clustering key? Massive rework could be required, especially if tables have to be redesigned/restructured. Even the simplest case of just changing the clustered index to something else requires lots of work.

You are right and I am wrong. There is absolutely NO REASON ON EARTH that you should EVER use an identity column as a clustered index. Happy now???

No, that's the same lie about me. I never stated one should never use an identity column. Contrarily, when I suggest anything but an identity column as a clustering key, you refer me to a video to remove my ignorance of the absolute necessity of using only identities as clustering keys.