I have a Tektronix 7834 analog storage oscilloscope for making 4 channel 300 MHz captures at 250,000+ sweeps per second (but not at the same time) over a span of seconds to minutes. Technically it can go for hours but it is already difficult enough to use and bistable storage mode is painful on the eyes. I have a pair of 2230s for making 4 channel captures over hours to days if necessary.
Neither of these have anything close to the storage capabilities of the DS1000Z or DS2000. The 7834 is limited (according to it's datasheet) to 30 minutes of time,
The Tektronix specifications are pessimistic (*) but storage time in bistable mode, which I would hate to use for the reason I gave, lasts essentially until power is removed and oddly enough generally even after that; that may seem a little odd but analog storage CRTs are weird. In reality I would never use persistence longer than a few minutes deliberately but it is handy for fast glitch hunting and as a variable persistence sampling display which is what I mostly use it for.
The DS2302A has almost the same bandwidth, the 7834 can be configured as fast as 400 MHz but the difference is not significant, but costs more than every oscilloscope I have combined without replacing the capabilities of all of them.
I would never recommend a 7834 to someone unless they had a specific requirement that it filled which modern but cheap DSOs did not and even then I would try to find an alternative. That is a very short list even if you include various ways a 7834 may be used to crush invaders.
and the 2230's can save a maximum of 3 waveforms (each @ 1k compressed). There is simply no way you can, for example, capture and store a 14k waveform once per second for 18 hours.
Or one channel with a 4k record length or 2 channels with 2k record lengths and half of those again if peak detection is used. They can actually store lots of waveform but that requires the memory backup option. These oscilloscopes also support external clocking for arbitrarily long duration recording but I have never needed it. If I am hunting for something specific over a long period of time, I would be using a qualified trigger circuit of some sort which Rigols are better at. If I am not sure of what I am looking for, I would have to rely on envelope detection and again, the Rigols would probably perform better for this but I am no longer sure.
None of them can capture and store a 14k waveform once per second for 18 hours but there are many things the Rigols cannot do that these can and they are the things I require. I keep being told that the Rigols are incapable of measuring the things I would like including for instance RMS jitter to high resolution or high resolution jitter at all. That is something I could use. I can do it now even within a 100 MHz bandwidth but it is not as convenient as it could be.
(*) Tektronix at the time was routinely pessimistic in their specifications even for marketing. They would generally list the minimum guaranteed number like 500 MHz for an oscilloscope which typically was more like 700 MHz or 600 picoseconds for pulse generator which actually performs more like 450 picoseconds. The 30 minute time for the 7834 variable persistence storage display probably reflects the amount of patience someone had watching it to make sure it met the specification they promised. The same statement could be made about transistor leakage or operational amplifier input bias current or noise. Testing takes time and costs money.