jonnify·bcs /bcs/overview/stack-and-floor

B0.3 · dive 3 of 3

The stack and the floor.

The shape of what the course builds: the volatile runtime tiers above — echo_data, echo_mq, echo_store — over the durable floor that makes state survive.

The stack over the durable floor. Select a layer.
BCS — your systems echo_data · branded ids, stores, CHAMPecho_mq · the Valkey-native busecho_store · the declared near-cache volatile above · durable below Echo Persistence — the durable floorETS L1Valkey L2CubDB · FjallTigris
readout · Select a layer. Volatile tiers above for speed; a durable floor below so state survives the heap. Reading top to bottom is reading the course.
§1

The idea

the stack · the floor

The course builds a stack, and the stack has a shape worth seeing before the chapters fill it in. At the top are your systems, written against BCS. Beneath them, echo_data provides the branded id, the property stores, and the CHAMP forest. Beneath that, echo_mq is the Valkey-native bus that moves work between systems. Then echo_store is the declared near-cache, holding hot state close to the systems that read it.

Those tiers are fast because they are volatile, and a system of record cannot be only volatile. The durable floor — Echo Persistence — sits beneath them: an ETS head, the Valkey bus and its L2, a durable local page tier on CubDB or Fjall, and a remote tier on Tigris. State lives somewhere fast and is made durable beneath, on a dial the system sets.

The line between the volatile tiers and the durable floor is the one the persistence chapter draws with care. Part I says state lives somewhere; Part II shows where in the heap; the floor chapter says where it lives when the heap is gone. Reading the stack top to bottom is reading the course in order.

§4

The doors

the sibling courses
§2

References

sources · related

Sources

Related