Originally VSAM supported RRDS (BDAM replacement), KSDS (ISAM replacement) and ESDS (Entry Sequenced Data Set), which is VSAM-speak for sequential file. They never really caught on that well, as they didn't offer any advantages over QSAM, and in the early days by the time you had defined the dataspace and the dataset within it they were more trouble than they were worth.
You can update records in place, but you can't delete them, although you can get around this by having a field on the record that identifies it as being logically deleted. Eventually at some stage you are supposed to copy the file and drop all the logically deleted records. But it's not very clean as it requires your application to know what the flag means.
Unless the file already exists and you are effectively stuck with it, you might find it easier to use QSAM...