Let me add a couple of (too late?) comments regarding the Rigol RSA3015E. First though, I might mention my approach to Chinese test equipment, given a background along the lines mentioned by lunix (above). In general, I now set a "no regrets" ceiling on the spend, such that if the equipment does only a few jobs before entering forced requirement I won't feel the need to for too much open weeping. In the case of spectrum analysers, that figure is probably around AUD5k. More than that and I start looking for deals on new or used R&S, Agilent etc.
On the RSA3015E-TG, I've found my unit to be useful for a several development projects and to perform acceptably, although I have the feeling that margins in some basic RF performance (including spurious responses) have been trimmed to accommodate the real-time system. That said, the RT side of things was a major reason for my purchase and the unit has done its job in the development of a suite of LF/MF Class D/E power amplifiers.
The bad news is Rigol support and customer engagement. My analyzer developed a crippling problem with the RT system in wideband mode (10 MHz), with hugely elevated noise floor and artifacts. The Australian distributor quickly confirmed my complaint and the unit was eventually fixed under warranty but not until THREE returns had taken place, including one to China and two to Sydney. TWO new motherboards were fitted, and along the way it became apparent that Rigol struggle with curation of customer and instrument databases, including the links between instruments and accessories such as preamps and specialized software modules. All in all, it was a great demonstration of left-right coordination issues, despite the best efforts of the Australian distributors who, by demonstration, also turned out to be far down the Rigol service food chain. The most honorable mention in the saga goes to Rigol North America who, despite having no financial interest, went out of their way to help, especially in terms of verifying my observations on their demonstration equipment.
That's not the end of it, though. I like to understand my test equipment and verify correct operation, at least in the modes of importance to me. I noted early on the step in the HF noise floor of the RS3015E (I think I am "some user" in the Rigol FAQ on this issue) and, while it's not a serious measurement impediment, it is inconvenient and remains unaddressed, or even intelligibly explained. In a second example, I noted severe shifts in displayed LF spectra when scans start <20 kHz and include RB's of < 1 kHz (if my memory is correct). This is clearly a firmware bug and remains unfixed, despite the fault being verified, logged and submitted to Rigol by Rigol North America. I could go on and mention display shifts in high resolution modes when locked to an external frequency reference, etc. The important point, though, is that firmware updates addressing reported problems have been non-existent and, perhaps more worryingly, the instruments were apparently released without much end-to-end checking by experienced RF engineers. The LF problem is especially egregious in that it includes the frequency range of considerable interest to EMC practitioners - key targets for the RSA range marketing material.
Most of the issues can be lived with when you know they're there but with a black box user mentality the RSA3015E (and maybe the rest of the RSA series, for all I know) would set more than the average number of user traps. My unit is still earning its keep but, in all honesty, I'd be hard pressed to recommend the model to anyone else. In some ways the smaller Rigol 800 series analyzers are a better capped investment but if you want better phase noise and RT capability, you'll be looking at the RSA range or equivalents from other manufacturers.