Oracles bring off-chain data (such as cryptocurrency prices, weather, etc.) onto the chain. This may seem simple, but in reality, there are two completely different approaches: one is active data pushing, and the other is contract on-demand pulling.

**Push-Oracles — Broadcast Mode**

This is the most popular method today, similar to Chainlink's Data Feeds system. Oracle nodes operate automatically based on preset conditions, such as pushing data every 5 minutes or immediately updating on-chain when the price fluctuates by more than 0.5%.

Sounds good? It indeed has two advantages — DApp developers don't need to bother with request logic and can directly read the ready-made on-chain data; users also get data quickly because it’s already waiting there.

But there’s no free lunch. The issue lies in costs: as long as conditions are met, each update consumes Gas, regardless of whether anyone actually uses that data. Imagine supporting 1000 types of RWA assets, pushing updates every few minutes — the Gas costs would be astronomical. For trading pairs with low volume, this mode is especially wasteful.

**Pull-Oracles — On-Demand Mode**

Projects like Tellor(TRB) are exploring another path: contracts only pull data when there is a demand. The benefit is clear — if no one uses it, no money is spent; it’s a pay-as-you-go model, offering much better scalability.

Each mode has its trade-offs; the key depends on what the application scenario requires.