Hold off and delayed timebase -to my knowledge- are also standard on any DSO.
More recent DSOs lack a delayed timebase. This is often obscured in the documentation which may refer to some type of delay function which actually only operates within the existing record and is what Tektronix used to call horizontal magnification. A good way to determine this is lack of trigger after delay capability.
Long record lengths are not a complete substitute for delayed sweep or acquisition. Try measuring the pulse to pulse jitter of a one pulse per second output from a GPS with a long record at the maximum sampling rate of a DSO. 1 Gsample of memory would be required to achieve 1 nanosecond resolution although some expensive DSOs have ways to do this without a long record length.
Agree with ferdieCX, its better to save the money, and look for a great value for money "NEW" DSO, even the mediocre ones are still much better than those old anchor boat DSO in term of bandwidth, capture rate and especially memory depth, cause now memory is very cheap compared to decades ago.
The issue here is that the old DSO being discussed has a bandwidth which would be very expensive in a new instrument. While equivalent time sampling rate can make up for real time sampling rate on periodic signals, nothing can make up for low bandwidth.
I really think there is a market for high bandwidth low sample rate DSO but nobody makes such a thing.
Worst all of them use customized chips that once hit, its not economically to fix anymore and become a burden.
This is not any different between new and old DSOs except in some unusual examples where complete documentation is provided. Any customized ICs or hybrids are usually the most reliable parts.
Old ETS DSOs which had high bandwidth but low sample rate were used in semiconductor characterization and automatic test applications where their low sample rate and high generation time were not disadvantages. They were not intended for single shot applications except at the lowest frequencies. They are still useful especially for their high bandwidth for general purpose development within their limitations. They often had more than 8 bit resolution. They usually *lack* peak detection which is what I consider their biggest disadvantage.
The Tektronix 7854 (400MHz 1MS/s 10bits), 11201A (400MHz 20MS/s 9bits), and 11400 (1GHz 20MS/s 10bits) are good examples of early DSOs like this.