We have in the region of 100 databases on servers at our ISP
The ISP has comprehensive backups to tape, but they take a long time to restore anything ... so we need to keep backups "online" for a reasonable length of time so we can recover from them first-and-foremost.
We produce Full backups once a week, and Differential backups daily. We keep Full backups for a month and Differential backups for seven days
We produce TLog backups every 10 minutes, but only store them (online) for two days.
So we can recover to point-in-time for the previous 2 days, to "midnight" for the last 7 days, and to "Sunday before" for the last 4 weeks.
The use of Differential backups reduces the total size of our backups, and has enabled us to keep more "elapsed time" online.
All backups are stored in NTFS compressed folders. We've never had a problem, but I don't think we have any databases > 30GB on these "shared DB servers"
During database maintenance (Index defrag etc.) we increase the frequency of our TLog backups to every minute. We found that the one thing which extended the DB size was the DB maintenance, so by doing this we have been able to reduce the operating size of our databases, and in turn free up some disk space for more backups. (Backups generally on different drives to MDF or LDF, but Batch operations copy older, stale, backup files to drives containing MDF / LDF just so we can keep them "online")
P.S. SQL 2008 has inbuilt compressed backup capability, in case you are planning to upgrade