init projectmycelium

specs/slices.md

@@ -0,0 +1,118 @@
+# Slices
+
+The fundamental unit of compute resource in the TF Marketplace
+
+Each slice has
+
+- compute e.g. 1 vcpu
+- memory (typically a multitude of 4 GB)
+- ssd storage capacity e.g. (200 GB)
+- min uptime (guaranteed by the provider, will lose TFP points if not achieved)
+- min bandiwdth (guaranteed by the provider)
+- passmark
+- reputation (as build up over time and measured by the ThreeFold system)
+- optional: public IP address (v6 or v4)
+- optional: certification flag (indicates the slice is provided by a [certified](./certified.md) farmer with ThreeFold support)
+- pricing in [TFP](./tfp.md) per hour/day/month/year (farmer can change the price)
+
+The farmer has lot of freedom to define these oaraneters
+
+## Farmers (Resource Providers)
+
+*   **Responsibilities:**
+    *   Register their hardware nodes (computers) with the platform.
+    *   Set pricing for their available capacity (Slices), guided by platform recommendations or within defined price bands.
+    *   Define and commit to Service Level Agreements (SLAs) for bandwidth and uptime.
+*   **Compensation:** Earn TFP based on the utilization of their registered capacity.
+*   **Monitoring:** Access to a UI to view their node status, earnings, and utilization metrics.
+*   **Risk:** Subject to "slashing" (loss of accrued TFP) if their nodes fail to meet committed SLAs.
+
+## Famer defines the slices
+
+The TF Marketplace
+
+- Users acquire slices through the [Marketplace](marketplace.md)
+- Slices are derived from physical nodes registered by "Farmers"
+- Example a node with
+  - 25,000 passmark
+  - 16 vcores
+  - 32GB RAM total capacity
+  - 2 * 1 GB SSD capacity
+  - no pub ip addresses
+- can deliver following slices
+  - 32 GB with 16 cores, 2 TB, 25.000 passmark
+  - 2 x 16 GB with 8 cores and 1 TB, 12500 passmark
+  - 4 x 8 GB with 4 cores and 0.5 TB
+  - 8 x 4GB with 2 cores and 0.25 TB per slice  (8 slices)
+  - farmer defined SLA
+    - 99.8%
+    - min 1 Mbit/sec
+  - farmer defines price
+    - e.g. hour price is 2x monthly price is 2x yearly price and is 10 TFP
+- The platform dynamically recalculates available slices on a node post-reservation
+  - e.g. a user buys 16 GB slice, means now only 1x 16 GB left, or 2x 8GB or 4x 4GB ... 
+
+## Management Features
+
+The user can do the following with a slice:
+
+- **Lifecycle Management**:
+  - OS deployment support for standard operating systems (Ubuntu, Alpine Linux, etc.)
+  - Start, stop, reboot capabilities
+- **Data Operations**:
+  - Backup functionality
+  - Restore capabilities
+- **Portability**:
+  - Export & import functionality e.g. to S3 or TF Quantum Safe Storage System 
+- **Ownership & co-administration rights**:
+  - Transferable between users
+  - Can be shared among multiple users
+
+## Marketplace Dynamics
+
+- **Pricing**:
+  - Farmers set prices for their node capacity/slices
+  - Pricing may follow platform-defined min/max brackets
+  - Platform may provide pricing recommendations
+- **Discovery**:
+  - Users browse available slices using filters:
+    - Geographic location
+    - Node reputation
+    - Uptime statistics
+    - Bandwidth availability
+    - Price ranges
+    - Certification status (see [Certified](./certified.md))
+- **Billing Models**:
+  - Flexible terms available:
+    - Per hour usage
+    - Per month subscription
+    - Per year commitment
+  - Discounts available for longer-term commitments
+- **Order System** (Future Enhancement):
+  - Buy/sell order book
+  - Users can place bids for VMs meeting specific criteria:
+    - Target price
+    - Preferred location
+    - Minimum reputation score
+
+## Networking
+
+- All slices are natively integrated with the "Mycelium" network
+- Connected to the broader TF Marketplace
+
+## Access States
+
+- **Full Access**:
+  - User has complete control over the slice
+  - Direct management of all aspects
+- **Managed Access**:
+  - User owns the slice
+  - Direct access might be restricted if a platform-managed "Solution" is running on it
+
+## Primary Use Cases
+
+- Running general-purpose VMs
+- Running AI workloads
+- Hosting specialized executors (e.g., "Heroscript executor" - planned for Phase 2)
+- Serving as the underlying infrastructure for "Solutions"
+- Supporting serverless functions