The short answer is that rETH liquidity comes from a mix of market trading and protocol-side ETH becoming available again through Rocket Pool's internal mechanics. Rocket Pool documentation around Atlas and later governance work shows that returned ETH can flow back into the deposit pool and support unstaking at the protocol rate, but timely withdrawal liquidity has also been important enough to motivate later RPIPs. So rETH redemption is not just 'burn token, get ETH instantly' in every situation; it depends on available protocol liquidity and, in the future, potentially on more explicit withdrawal-liquidity mechanisms.
Where redemption liquidity comes from today
Rocket Pool's Atlas documentation explains that when skimmed rewards are distributed, the rETH holders' share goes back into the deposit pool. That matters because the deposit pool is one of the places protocol-side ETH can become available for unstaking at the protocol rate.
In other words, redemption liquidity is partly a function of normal protocol cash flow rather than only of a separate dedicated withdrawal system.
Why timing can still be uneven
RPIP-74 explicitly treats support for redemptions as a governance priority when discussing deposit-pool usage. That is a sign that redemption liquidity is important enough to influence other protocol decisions such as queue behavior and where ETH should be directed.
So even before talking about new withdrawal systems, the protocol already treats rETH exit liquidity as something that depends on broader ETH allocation choices.
What draft governance work is trying to change
Draft RPIP-71 goes further and describes a desired future where pool stakers can signal a wish to unstake at the protocol rate and have validator exits help satisfy that demand. The proposal is explicit that this is a draft and not yet a settled implemented standard.
That makes RPIP-71 useful as direction-of-travel evidence, not as proof that the full user-signaled withdrawal system already exists in final form.
What this means for users
If someone wants the practical answer, they should think of rETH exit liquidity as a live protocol-liquidity question rather than a guaranteed instant redemption promise in every state of the system.
If they want the governance answer, they should know Rocket Pool is actively thinking about improving that experience, but the most ambitious withdrawal-liquidity design work is still draft-stage rather than final protocol law.