/miasma

Cloud modeling library

Primary LanguageRubyOtherNOASSERTION

 ███▄ ▄███▓ ██▓ ▄▄▄        ██████  ███▄ ▄███▓ ▄▄▄
▓██▒▀█▀ ██▒▓██▒▒████▄    ▒██    ▒ ▓██▒▀█▀ ██▒▒████▄
▓██    ▓██░▒██▒▒██  ▀█▄  ░ ▓██▄   ▓██    ▓██░▒██  ▀█▄
▒██    ▒██ ░██░░██▄▄▄▄██   ▒   ██▒▒██    ▒██ ░██▄▄▄▄██
▒██▒   ░██▒░██░ ▓█   ▓██▒▒██████▒▒▒██▒   ░██▒ ▓█   ▓██▒
░ ▒░   ░  ░░▓   ▒▒   ▓▒█░▒ ▒▓▒ ▒ ░░ ▒░   ░  ░ ▒▒   ▓▒█░
░  ░      ░ ▒ ░  ▒   ▒▒ ░░ ░▒  ░ ░░  ░      ░  ▒   ▒▒ ░
░      ░    ▒ ░  ░   ▒   ░  ░  ░  ░      ░     ░   ▒
       ░    ░        ░  ░      ░         ░         ░  ░

Overview

Miasma is YACACL (Yet Another Cloud API Client Library). Instead of attempting to cover all the functionalities of all the different cloud and virt APIs, miasma is focused on providing a common modeling that will "just work". This means there may be many things that seem to be missing, but that's probably okay. Consistency is more important than overall completeness. Miasma isn't trying to be a replacement for libraries like fog, rather it's attempting to supplement those libraries by providing a consistent modeling API.

Usage

Example

Lets have a look at using the compute model:

compute = Miasma.api(
  :type => :compute,
  :provider => :aws,
  :credentials => {
    :aws_secret_access_key => 'SECRET',
    :aws_access_key_id => 'KEY_ID',
    :aws_region => 'us-west-2'
  }
)

With this we can now list existing servers:

compute.servers.all

This will provide an array of Miasma::Models::Compute::Server instances. It will also cache the result so subsequent calls will not require another API call. The list can be refreshed by requesting a reload:

compute.servers.reload

Switching providers

Switching providers requires modification to the API parameters:

compute = Miasma.api(
  :provider => :rackspace,
  :credentials => {
    :rackspace_username => 'USER',
    :rackspace_api_key => 'KEY',
    :rackspace_region => 'REGION'
  }
)

The compute API will act exactly the same as before, now using Rackspace instead of AWS.

Design Objectives

This library is following a few simple objectives:

  • Light weight
  • Consistent API

The availabile API is defined first via the models, then concrete implementations are built via available provider interfaces. This means the provider code is structured to support the models instead of the models being built around specific providers. The result is a clean model interface providing consistency regardless of the provider backend.

The "weight" of the library is kept light by using a few simple approaches. All code is lazy loaded, so nothing will be loaded into the runtime until it is actually required. Dependencies are also kept very light, to reduce the number of required libraries needing to be loaded. Parser wrapping libraries are also used (multi_json and multi_xml) allowing a choice of actual parsing backends in use. This removes dependencies on nokogiri unless it's actually desired and increases installation speeds.

Features

Thin Models

Thin models are a stripped down model that provides a subset of information that an actual instance of the model may contain. For instance, an AutoScale::Group may contain a list of Compute::Servers. The collection provided within the AutoScale::Group will be created via the resulting API information on the group itself. However, since we can provide expected mappings to what API provides Compute and know these instances will be within the servers collection, we can use the #expand helper to automatically load the full instance:

auto_scale = Miasma.api(:type => :auto_scale, :provider ...)
group = auto_scale.groups.first

# this list will provide the `ThinModel` instances:
p group.servers.all

# this list will provide the full instances:
p group.servers.all.map(&:expand)

Automatic API switching

Resources within a specific provider can span multiple API endpoints. To deal with this, the provider API implemenetations provide an #api_for method which will automatically build a new API instance. This allows Miasma to hop APIs under the hood to expand ThinModels as shown above.

Current status

Miasma is currently under active development and is in a beta state. Models are still being implemented and spec coverage is following closely behind the model completions.

Currently Supported Providers

Coverage currently varies from provider to provider based on functionality restrictions and in progress implementation goals. The README on each library should provide a simple feature matrix for a quick check on support availability:

Info