I have a field that is 95% hebrew and 5% arabic. The table is  10,000,000
recs, so for storage purpose I prefer not to use unicode. Is there a
collation that contains both?

Gal,

I don't believe that you can be successful with this.  If you look at the
following site, you can compare Hebrew and Arabic byte code and unicode
values.  You will see that in non-Unicode the same byte-code can mean two
different characters, depending on the language.
http://www.collation-charts.org/mssql/

If you feel like you can manage the display issues based on information that
you have, then great.  However, the single-byte collations will not help you
with this.

That is why Unicode exists, to provide support for this type of thing.

RLF

Show quoteHide quote

You can create you own codepage, separate this table into two tables or add
a field to make a distinction between these two languages.

However, the choice of using Unicode versus the use of 8 bit codepage goes
far beyond than the simple decision about saving some hard drive space.
Even if your database was 100% hebrew or 100% arabic, without any mix
between these two languages, it would still be a good idea to choose Unicode
over a 8 bit codepage; otherwise, you might hit later some serious blocks
down the road.  The percentage of hard drive space that you'll save today
might cost you big time tomorrow; so you should be pretty sure about the
need of saving some hard drive space before going into the direction of
using any 8 bit character encoding.

Show quoteHide quote

in full recovery mode but log keeps self-truncating
Server Disk Space
Max server memory being ignored
Security Question
Consolidate datafiles to one drive
Cumulative package 2 for SQL 2008
Maintenence Cleanup Task does not delete backup files
How to do bulk update in SQL Server 2005.
Taking database offline taking long time
32 vs 64 bit SQL Server hotfixes