Dell Dude (1/29/2015)

---

ScottPletcher (1/29/2015)

---

So I guess the table name is "elmah"?! It's somewhat confusing with all the different "table names" given.

DROP INDEX PK_ELMAH_Error ON...

True,...

So I guess the table name is "elmah"?! It's somewhat confusing with all the different "table names" given.

DROP INDEX PK_ELMAH_Error ON elmah

DROP INDEX idx_ELMAH_TimeUtc...

Btw, be sure to script out the existing indexes before you delete them so you can recreate them easily ;-).

Dell Dude (1/29/2015)

---

ScottPletcher (1/29/2015)

---

Drop all the indexes, starting with the nonclustered, then recreate them, starting with clustered. Unless the table itself is damaged, that's likely your best chance to...

You could try a nonclustered index on a that contains id,

and a nonclustered index on b that contains id.

That may or may not help, depending on the total sizes of...

Drop all the indexes, starting with the nonclustered, then recreate them, starting with clustered. Unless the table itself is damaged, that's likely your best chance to clean this up.

I believe CROSS APPLY would be more appropriate to the original requirement ... but I just see that may be a moot point now :-D.

SELECT

...

I believe either ProductId or productId are...

Steve Jones - SSC Editor (1/29/2015)

---

Eric M Russell (1/27/2015)

---

Regardless of datatype, I don't want to see a column names having no contextual prefix, meaning just "ID" by itself, on any...

Based solely on the existing query -- having not yet reviewed index missing stats and index usage stats for this table -- you could add a clustered index on:

si.ShippingKey

Nonclustered...

TomThomson (1/28/2015)

---

Eric M Russell (1/27/2015)

---

andrew gothard (1/27/2015)

---

Eric M Russell (1/21/2015)

---

ProductID, ProductId, and product_id are all respectable options, so long as it's done consistently.

However, I would take issue with ID...

Something like this:

SELECT

SUM(CASE WHEN ht.asAtDate >= DATEADD(DAY, -7, today_midnight)

THEN ht.[numLow] + ht.[numMedium]...

Is ProdId a guaranteed unique value in ProductsL1? The error message indicates that the trigger might be trying to UPDATE more than one row.

I'd stick with the AFTER trigger. Your initial trigger was pretty much spot on.

In case you create other AFTER triggers later, you might want to explicitly designate that trigger...