How and where can I learn more about financial data feeds? I'm specifically interested in learning how feeds are implemented, i.e. are feeds implemented as SOAP/RESTful Web Services? Or if they are implemented differently, how so and as what?
I have actually worked with a couple of data feeds and each and every one has been different. The only thing they had in common was that none of them used SOAP/RESTful Web Services or any of that other fancy message exchange stuff. Since these feeds often transport a lot of data, they are optimized for minimal bandwidth usage. A bunch of XML tags would just mean unnecessary overhead.
In the Reuters Market Data System protocol, for instance, a message is essentially a string with a bunch of special characters introducing the header, field identifiers and other control sequences (e.g. to indicate a partial update of a field value, again to skimp on the number of bytes that have to be transmitted). Another provider I have worked with, Invision, used something similar to that, although not exactly like it. And some providers might even roll their own. I have, for instance, seen one that just transmitted three semicolon-delimited values: the ISIN of the instrument the message pertained to, a bid and an ask quote.
Of course there is FIX, which is supposedly a standard, but even though many providers and exchanges claim to support it, if you dig a little deeper, you will find that they are stretching the definition of the standardized fields or simply put their stuff in custom fields that aren't standardized. Also, I have seen FIX used primarily in order-management scenarios and only once for a price feed. Again, its overhead is probably a deterrent, while its advantages such as recovering messages after a connection is re-established is not as relevant for real-time price feeds.
Unfortunately, as far as I can tell, the specifications for such data feeds are not public and you have to be a paying customer to get your hands on some documentation.
This is a big space, and the technologies are entirely dependent on the provider. Reuters have different technologies, so does Bloomberg, and then take every exchange and provider in between - they mostly have different technologies.
I think a better approach for you may be to learn some common open formats (which may be in use by multiple exchanges/providers), for example FIX is a good starting place. Again this protocol is very detailed, but should give you an idea of how market data is disseminated and the order flow typically works (again exchanges are free to implement the subset of the FIX protocol they require).
Once you've got to grips with that, try to understand some of the derivations such as FAST (Fix adapted for streaming). These two technologies are quite common, beyond that if you want to understand a specific technology, you need to get the specs from them and I'm not sure if this is possible or publicly available.