> ## Documentation Index
> Fetch the complete documentation index at: https://lattora.puneetchandna.com/llms.txt
> Use this file to discover all available pages before exploring further.

# Capabilities and migration

> Current scope, saved-data compatibility and planned work.

Lattora 2.1.0 is a reproducible replacement for the legacy research project.
It restores a narrower set of capabilities with explicit contracts.

## Current scope

| Capability | 2.1.0 behavior |
| - | - |
| Algorithms | HO, GA, FirstFit and BestFit with defined variants, pairing and budgets |
| Simulation | CloudSim Plus 8.5.7; static placement and strict reservations |
| Frozen profiles | 4/40/450 cases; protocol/schema 2; maximum 100 VMs |
| Configurable campaigns | Descriptive Stress schema 1; size, N/T/R and seed controls |
| Concurrency | CPU/shared-heap bounded independent cases; sequential event loops |
| Interface | Terminal workbench, real progress, retained history and validation |
| Distribution | Linux x86-64/ARM64 and Apple Silicon; bundled runtimes; offline experiments |
| Recovery | Manual update/rollback; results and settings retained |
| Resume | Unavailable; failed fragments cannot be pooled |

## Migrate carefully

Use `lattora` for the installed app and `./lattora.sh` for source development.
`./cloudsim.sh` remains a compatibility alias. Source tools/settings remain under `.cloudsim/`.
Java package `hiippo` and existing scientific identifiers keep their names.

Legacy mutable properties, runners and policy/statistical classes are outside the supported path.
Old `Config.properties` and algorithm controls are not current APIs.
Pre-v2 results remain invalid research evidence. Keep them for historical diagnosis;
there is no automatic conversion to accepted v2 data.

The power objective is estimated W. Measured energy is J/kWh; SLA uses cloudlet violations.
These semantics differ from legacy multi-objective claims.
Read [the protocol](/concepts/research-protocol) before comparing old and new results.

## Planned capabilities

These issues were open when this 2.1.0 edition was checked.
Existing Stress/TUI foundations do not complete their broader requirements.

| Issue | Planned extension |
| - | - |
| [#16](https://github.com/puneet-chandna/Lattora/issues/16) | Versioned workload/hardware scenario families |
| [#17](https://github.com/puneet-chandna/Lattora/issues/17) | Power, utilization, SLA and weighted objective selection |
| [#18](https://github.com/puneet-chandna/Lattora/issues/18) | Rich Python analysis, plots, reports and optional LLM narration |
| [#19](https://github.com/puneet-chandna/Lattora/issues/19) | Stronger baselines and verified tractable optima |
| [#20](https://github.com/puneet-chandna/Lattora/issues/20) | Reproducible tuning with separate held-out evaluation |
| [#21](https://github.com/puneet-chandna/Lattora/issues/21) | Full larger/sustained campaigns and measured operating limits |
| [#22 to #27](https://github.com/puneet-chandna/Lattora/issues/22) | Graphical researcher workspace, shared algorithm contract and reviewed onboarding |
| [#13](https://github.com/puneet-chandna/Lattora/issues/13), [#14](https://github.com/puneet-chandna/Lattora/issues/14) | Dynamic/migration models and empirical calibration |

GPU, Pareto optimization, overcommit and host transition energy are outside the current model.
[Extension guidance](/guides/extending) describes the supported source boundary.

The [exact-tag release checks](https://github.com/puneet-chandna/Lattora/actions/runs/37265556152)
passed frozen profiles and bounded Stress on all three native targets.
They do not establish full Large/XLarge limits or production savings.


This documentation is built and hosted on [Mintlify](https://mintlify.com), a developer documentation platform.