with(nolock) or (nolock) - Is there a difference?

问题:

Everything is based on the assumption that with(nolock) is entirely appropriate for the situtation. There are already plenty of questions out there debating whether or not to use with(nolock).

I've looked around and haven't been able to find if there is an actual difference between using with(nolock):

select customer, zipcode from customers c with(nolock) 

or just (nolock):

select customer, zipcode from customers c (nolock) 

Is there a functional difference between the two? Stylistic?
Is one older than the other and has a chance of being deprecated?

回答1:

There is no functional difference, but eventually the syntax without WITH will not work. This has been deprecated:

select customer, zipcode from customers c (nolock) 

So you should be using this format:

select customer, zipcode from customers c with (nolock) 

Not using the WITH keyword for table hints has been deprecated since at least SQL Server 2008. Search the following topic for the phrase Specifying table hints without using the WITH keyword.:

http://msdn.microsoft.com/en-us/library/ms143729%28SQL.100%29.aspx

(Discussions about whether you should be using nolock at all, of course, are separate. I've blogged about them here.)

回答2:

Though we dont find difference between (nolock) and with(nolock) ... with (nolock) would not work in SQL Server 2000 version.

And I also noticed that when you try to pull data from linked servers, just ' (nolock) ' will not work whereas you should use ' with (nolock) '.

-- this will not work 
select * from server1.DB1.dbo.table1 (nolock)

-- this will  work 
select * from server1.DB1.dbo.table1 with (nolock)

回答3:

It really depends on which version of SQL Server you're on.

Checking out the latest documentation for SQL Server 2012 table hints omitting WITH is a deprecated feature. So while from customers c (nolock) will probably work; you should really be using from customers c WITH (nolock)

Note that this is different than from customers nolock; where nolock would serve as the table alias.

Functionally; they appear to be the same.

回答4:

I tried this for a 170000+ data row result, however I did not see any difference through the query execution plan. Both work in the same way.