sql server - Does an index already cover a clustered primary key? -

let's have table this:

create table t(   [guid] [uniqueidentifier] not null,   [category] [nvarchar](400)   {,...other columns}   ) 

where guid primary key, , has clustered index.

now, want index covers both category , guid, because i'm rolling other stuff related t category, , want avoid including t table itself.

is sufficient create index covering category, or need include guid well?

i expect sql server indexes point directly page offsets in t rather referring guid primary key value, means would need explicitly include pk column avoid hitting t. case?

actually assumption wrong - sql server non-clustered indices include clustering key (single or multiple columns) , not point directly @ physical page.

this prevents sql server having reorganize , update lots of index entries when page needs split in 2 or relocated. if seeking in non-clustered index , find value, have clustering key , sql server need "bookmark lookup" (or key lookup) retrieve actual data page (the leaf page in clustering index) whole set of data belonging single row.

that said - if ever have situation depends on ordering of key columns, still might need create index on (guid, category) - of course, in case, sql server smart enough figure out clustering key column in index , won't adding 1 more time.

the fact clustering key column(s) inlcuded in every single non-clustered index strong reason why clustering keys should narrow, static , unique. making them wide (anything beyond 8 byte) sure recipe bloat , slow-down.