Basically, when you're on service you are billing the patient for the services you provide. After all the back and forth between your hospital/provider organization and the insurance company/collection agency/whatever, the hospital/provider organization gets paid for the services you provided. Each billing code is associated with a number of units of clinical "work", which is the RVU. In a perfect world, the (number of RVUs) x (reimbursement rate) = (money received from billing) - (overhead) and everything is balanced. But for a number of reasons that might not be the case. For instance, a big university might want to have some behavioral neurologists around, but those people need to make a living wage. So you might boost their RVU reimbursement rate so that the amount you're paying them for a new patient clinic visit exceeds what their billing brings in. Then you balance the books by paying your neuroICU providers less per RVU for the critical care time they bill for. In a big university, people tend to understand that this is necessary for the common good, but in a private hospital maybe not. It can even happen at a grander scale, where the entire dermatology department might make less per RVU so they can keep the psychiatry department open.
The number of RVUs you can generate will vary a lot just based on the patient load, new vs. follow-up breakdown, and time on service, so you'll need to do a little local homework to get a feel for what you're likely to generate per week on the inpatient service, per week on the consult service, etc.
The reimbursement per RVU varies dramatically by setting, department, ICU vs inpatient vs. outpatient, etc. For instance, where I work, RVUs generated in the ICU are reimbursed at a lower rate than inpatient RVUs, and inpatient RVUs are reimbursed at a lower rate than outpatient RVUs. Basically, the highest billers subsidize the rest of the department, and the fiscal focus is placed on maximizing outpatient clinic output because the hospital has made that a priority for patient access. In departments with a more procedural or surgery focus, RVUs can be determined by volume, seniority, or other factors like call burden. When you see academics making $150,000 a year, and private practice people making $300,000 a year, some of that is determined by volume (number of RVUs generated), and some of that is determined by the reimbursement rate.
Finally, there is the added issue of "hard money", which can be used to create a salary beyond the provider's clinical billing. For some departments, RVUs are reimbursed at a pretty low rate, with the rest of the money going into a big pool of money in the department. This pool is then used to support guaranteed salaries for the providers, with a bonus structure to reward excess billing beyond base expectations. Philosophies on these arrangements vary, with some preferring the transparency of the "eat what you kill" model, and others preferring the somewhat greater stability of a guaranteed salary plus bonuses for reaching certain RVU thresholds.