That sounds dangerous because you just can't be naive when trying to manage
a table that grows at 1 billion rows per year. In any event, one of the most
effective technique in managing large databases is partition. Or at an even
higher level, the approach is to reduce--as much as possible--the unit of
operations that threaten your service availability.

Linchi

Attention to detail (especially indexing) is the main key to performance.
With an efficient path to the data, queries should perform well regardless
of table size.

Manageability is a major consideration with large tables.  For example,
you'll need to maintain about 120% free space in order to rebuild the
clustered index.  Table partitioning can help in this regard and improve
performance of certain types of scans too.  Also SQL 2008 will introduce
compression features to reduce space requirements and improve performance
and manageability.  Note that these features are available only in
Enterprise and Developer editions.

I'll just echo the good advice you've already been given, but add a
few more points.  Just FYI, something like a million rows is small
these days and would require only modest care, but a billion, yes,
you're getting up there, that's a thousand times what modest care can
handle!  

(oh, and that's an American billion, 1e9, I hope, if it's a British
1e12, you're in big trouble! let's assume you meant 1,000,000,000)

A billion rows for real-time reporting?  Come now.  I presume only a
few of those rows will be current at any given time.  On any system
you'll want to do a lot of preprocessing of the data to date, in order
that real-time reports (eg, dashboards) can be done quickly.

Are these big rows?  How much disk space are you going to need, for
starters?  Remember, database storage adds overhead.

Are you sure it's a billion rows?  Maybe it's 365 rows with a ten
megabyte blob in each.

How are they arriving?  If in real time evenly 24x7, that's about 2k
rows/minute, say 30/second.  If they arrive in big files 10,000,000 at
a time to import, that's still an issue.

What's the application?  If this is something common, you might find a
solution is available off the shelf.

Josh