Best way to change primary key of a huge table in SQL Server - sql-server

Question

I have a table in a database with 580 million rows in it, and an awkward composite primary key. I would like to change the structure of the table to have an identity column as the primary key.
I am looking for some suggestions on the best possible way of doing this in the shortest amount of time.
We are using SQLServer 2008.
Current table structure:
CREATE TABLE [dbo].[POINTS_EARNED](
[CARD_ID] [int] NOT NULL,
[CYCLE_ID] [int] NOT NULL,
[POINTS_CODE] [int] NOT NULL,
[NO_POINTS] [int] NULL,
[ACCOUNT_ID] [int] NOT NULL,
[CREATED_DATE] [datetime] NULL,
[CREATED_BY] [varchar](20) NULL,
[LAST_MODIFIED_DATE] [datetime] NULL,
[LAST_MODIFIED_BY] [varchar](20) NULL,
[DELETED] [bit] NULL,
CONSTRAINT [PK_POINTS_EARNED] PRIMARY KEY CLUSTERED
(
[CARD_ID] ASC,
[CYCLE_ID] ASC,
[POINTS_CODE] ASC,
[ACCOUNT_ID] ASC
)WITH (PAD_INDEX = OFF, STATISTICS_NORECOMPUTE = OFF, IGNORE_DUP_KEY = OFF, ALLOW_ROW_LOCKS = ON, ALLOW_PAGE_LOCKS = ON, FILLFACTOR = 90) ON [PRIMARY]
) ON [PRIMARY]
New structure
CREATE TABLE [dbo].[POINTS_EARNED](
[Points_EARNED_ID] [int] [Identity] Primary Key,
[CARD_ID] [int] NOT NULL,
[CYCLE_ID] [int] NOT NULL,
[POINTS_CODE] [int] NOT NULL,
[NO_POINTS] [int] NULL,
[ACCOUNT_ID] [int] NOT NULL,
[CREATED_DATE] [datetime] NULL,
[CREATED_BY] [varchar](20) NULL,
[LAST_MODIFIED_DATE] [datetime] NULL,
[LAST_MODIFIED_BY] [varchar](20) NULL,
[DELETED] [bit] NULL
)WITH (PAD_INDEX = OFF, STATISTICS_NORECOMPUTE = OFF, IGNORE_DUP_KEY = OFF, ALLOW_ROW_LOCKS = ON, ALLOW_PAGE_LOCKS = ON, FILLFACTOR = 90) ON [PRIMARY]
) ON [PRIMARY]
In the past we have tried 2 ways of doing this on other tables that needed to be restructured:
Create an empty table in the new structure
INSERT INTO
Newtable SELECT * FROM Oldtable
rename old table oldtable_bak
rename new table as old table
Add indexes, etc to new table
Unfortunately, with large tables this tends to cause SSMS to crash, so we have copied the data by changing step 2 to be:
bcp data out of the old table into a text file, and then bcp it back into the new table from the text file, which seems to work, but it takes several hours.
I'm interested to know whether there is a better, more efficient way of doing this.

Answer 1

I'm not sure if it is more efficient, but the following is "better" in a certain sense:
Insert existing data into a new table
Truncate existing table
Alter table to had identity primary key
Insert into new table
The reason this is better is because it preserves triggers, constraints, permissions, and other references to the table. That can be quite handy many applications.
As your implicitly point out, you might want to remove all indexes from the original table and add them after the data is re-inserted. Generally, building indexes all at once is more efficient.