Presentation is loading. Please wait.

Presentation is loading. Please wait.

TOSCA Orchestration and Management in OpenStack

Similar presentations


Presentation on theme: "TOSCA Orchestration and Management in OpenStack"— Presentation transcript:

0 Enterprise Cloud Application Orchestration in OpenStack at UBS
Session: 4966 Enterprise Cloud Application Orchestration in OpenStack at UBS with Governance Policies Thursday, Feb. 25th, 9:30 AM – 10:15 AM Mandalay Bay SOUTH, Breakers L

1 TOSCA Orchestration and Management in OpenStack
Clusters, Policies, Network Function Virtualization (NFV)

2 Review: TOSCA’s Key Modeling Concepts that UBS found Interesting
Primary Goal was to Simplify Application-Centric Modeling for Cloud Describe Topology of an Applications’ “desired state” to a Cloud Service Provider TOSCA is an “Intent Model” that: Goes beyond traditional Information Models (IMs) Carries Configuration information described by Data Models (DMs) Promote Composition of Application Services for (Cloud) Platform providers Describe application’s (abstract) Requirements that can be fulfilled by other services’ Capabilities at design-time or run-time by: Linking to other TOSCA Service Templates via Service Catalogs <or> Dynamic fulfillment by Cloud Service (platform) Providers (e.g., IaaS, PaaS) Enable granular, operational Lifecycle management of the entire application topology or any of its composite services (including update-in-place) Support DevOps by associating different Repositories, Policies, Artifacts, Configurations (data) Update-in-place of individual (micro)services and/or sub-topologies. Encourage Portability via normative Node (resource) and Relationship types Growing set of Normative IaaS and PaaS types in the base specification Encourage Profile specifications for adding “domain-specific” type systems Move away from describing exactly HOW everything should be done Permit service providers to match their platform service Capabilities that best suits the applications requirements

3 TOSCA in Open Source & Open Standards
Heat-Translator Senlin Tacker TOSCA Template Validation and Translation to other DSLs Clustering / Policy with Monitoring Virt. Network Function (NFV) Manager (Software-Defined Networks) TOSCA Blueprint,Type & Workflow Designer alien4cloud Service Orchestration & Management Seaclouds Open, Multi-Cloud Managemnt CERN Indigo-DataCloud Data/computing platform targeted at scientific communities OPNFV Parser Deployment Template Translation TOSCA is technology agnostic, OpenStack was chosen place for a reference impl. works with traditional scripting solutions Ansible, Chef, Puppet, etc. The TOSCA impl. in OS is Heat Translator in OpenStack has support for OpenStack APIs today, but is able to translate and support other APIs as well in fact, the Translator and Parser are reusable libraries in Pypi that can support plug-ins for new APIs The TOSCA Service Templates and types can be realized with ANY target platform API by a TOSCA Orchestrator (either directly or directly by translation as we do in OpenStack via Heat)

4 OpenStack Update: New Services utilizing TOSCA in OpenStack
Already able to model & orchestrate TOSCA IaaS and PaaS patterns in OpenStack using the Heat Orchestration Service and the TOSCA Heat-Translator Logical PaaS Patterns via Heat / Heat-Translator IBM Bluemix Software Services Application Support Patterns App App Catalog (Murano) Container (Magnum) Cluster-Policy (Senlin) NFV (Tacker) IaaS Patterns via Heat / Heat-Translator Shared Image (Glance) Database (Trove) Identity (Keystone) Metering (Ceilometer) Core Compute (Nova) Storage (Cinder, Swift) Network (Neutron) Baremetal (Ironic) Physical Software-Defined Systems and Hardware Patterns IaaS, PaaS and SaaS is supported using the TOSCA Heat-Translator and the Heat Orchestration services New OpenStack Services moving to use TOSCA for modeling and orchestration: Senlin (Cluster-Policy) & Tacker (Network Function Virtualization)

5 OpenStack Senlin Project
Cluster – Policy Service

6 Senlin – Clustering and Policy service for OpenStack Clouds
Profiles – describe supported Container types and how to deploy them using their corresponding OpenStack services’ APIS; including: Linux Containers (Magnum) e.g., Docker, Stacks (Heat), VMs (Nova), Bare-metal (Ironic) Policies - framework to monitor and enforce various Operational policies on Cluster TOSCA Service Templates Senlin Policies (TOSCA) Supported Senlin Profiles Placement Scaling Load Balancing Deletion Performance Senlin Engine Linux (Magnum) Stacks (Heat) Load Profile + VMs (Nova) Bare-metal (Ironic) Deploy using APIs Topology of Cluster / Containers Register Alarms from Policies Monitoring (Ceilometer) Senlin plans to use TOSCA Policy definitions which can be associated with Clusters and Containers described by TOSCA Service Templates

7 TOSCA Attaching Operational Policies to Applications
Normative Operational Policies for: Placement (Affinity), Scaling and Performance with Rules that are evaluated to execute Automatic and Imperative Triggers 2 Policy Event (Type) Condition (Rule) Action (Trigger) my_cluster 1 Policy Event (Type) Condition (Rule) Action (Trigger) app_service Compute Capabilities Container ... Lifecycle create configure backend_app Compute 3 Policy Event (Type) Condition (Rule) Action (Trigger) web-app Compute my_database Compute Policies can be attached to: Interfaces for specific Operations to monitor lifecycle events on a resource (service) Nodes – facilitates resource (service) level monitoring Groups – facilitates composite service / application monitoring Abstract A key feature of any Cloud infrastructure is to provide auditing capabilities for compliance with security, operational and business processes. In this talk we provide an overview of the recent enhancements made in OpenStack projects to support API and security auditing using the DMTF Cloud Auditing Data Federation (CADF) standard. We will describe how auditing is seamlessly enabled for Nova, Glance, Swift, Cinder, Neutron and Keystone and illustrate what is audited, where it is stored, what the records contain and how this supports compliance. We will finish by presenting some possible future directions such as extending the use of CADF beyond audit to facilitate event correlation and federation across multiple tiers. Enables monitoring of Utilization, Health and Scaling at any level 7 7

8 1..N Triggers can be declared
<policy_name>: type: <policy_type_name> description: <policy_description> properties: <property_definitions> targets: [ <list_of_valid_target_resources> ] triggers: <trigger_symbolic_name_1>: event_type: <event_type_name> schedule: start_time: <iso 8601 date-time> end_time: <iso 8601 date-time> # Identify node (resource) to monitor target_filter: node: <node_template_name> | <node_type> # (optional) reference to a related node # via a requirement requirement: <requirement_name> # (optional) Capability within node to monitor capability: <capability_name> # Describes an attribute-relative test that # causes the trigger’s action to be invoked. condition: <constraint_clause> action: # implementation-specific operation name <operation_name>: inputs: <list_of_parameters> implementation: <script> | <service_name> 1..N Triggers can be declared TOSCA Policy Definition: Event Normative TOSCA Event Type That describes an event based upon a Resource “state” change. Or a change in one or more of the resources attribute value. Optional schedule Condition Identifies: the resource (Node) in the TOSCA model to monitor. Optionally, identify a Capability of the identified node. Describe the attribute (state) of the resource to evaluate (condition) Action Describes: An Operation (name) to invoke when the condition is met within the declared Implementation Optionally, pass in Input parameters to the operation along with any well-defined strategy values. 8 8

9 Senlin Example : TOSCA Topology for Kubernetes Pod, Docker Containers
Kubernetes “redis” example: TOSCA Service Template Cluster Scaling Policy Event = tosca.resource.utilization Condition = utilization > 80 % Action = Senlin.scale() redis-master-pod Cluster.Kubernetes ... capabilities: loadBalancer: properties: ... members: [ master-container, sentinel-container ] Kubernetes Pod modeled as a TOSCA Cluster (Group Type) built-in LoadBalancer Capability “Attach” Container Scaling Policy Event = tosca.resource.utilzation Condition = utilization > 90 % Action = Senlin.scale() Associated via TOSCA “MemberOf” Relationships (implied) “Attach” Docker Containers modeled as TOSCA Container Node Types master-container sentinel-container Container.App.Kubernetes Container.App.Kubernetes There was some assumptions made that the “cpu” was related to CGROUPS or “Control Groups” within Linux… from Wikipedia… One of the design goals of cgroups is to provide a unified interface to many different use cases, from controlling single processes (by using nice, for example) to whole operating system-level virtualization (as provided by OpenVZ, Linux-VServer or LXC, for example). Cgroups provides: Resource limitation: groups can be set to not exceed a configured memory limit, which also includes the file system cache[6][7] Prioritization: some groups may get a larger share of CPU utilization[8] or disk I/O throughput[9] Accounting: measures how much resources certain systems use, which may be used, for example, for billing purposes[10] Control: freezing the groups of processes, their checkpointing and restarting[10] Senlin scaling cluster / containers independently with Magnum service OpenStack Tokyo 9 9

10 OpenStack Tacker Project
Network Function Virtualization (NFV) Orchestrator / Manager Service for use in Cable and Telco industries and any Cloud network provider

11 OpenStack Heat Translator
OpenSource related to ETSI NFV and OASIS TOSCA TOSCA Service Templates OPNFV Parser OpenStack Heat Translator TOSCA NFV Models NSD VNFD VNFFGD Tacker NFV Orchestrator ETSI NFV Descriptor tosca-parser reference implements ETSI NFV Standards: NFVO / VNFM / Catalog TOSCA Policies heat-translator Cluster-Policy (Senlin) Orchestrate using Heat or Nova APIs

12 Why Network Function Virtualization (NFV) uses TOSCA
TOSCA Service Templates can describe these various levels of composition NFV Information model has such a layered structure: Network Services (NSDs) are composed of Virtual Network Functions (VNFDs), Virtual Links (VLDs), and sometimes actual Physical Network Functions (PNFDs), etc. VNFDs are composed of Virtualization Deployment Units (VDUs), with VLDs, etc. NSD VNFD VNFD VLD Any node in a TOSCA topology can be an abstraction of another layer or sub-topology VNFD VDU VLD VNFD Using the TOSCA, NFV multi-layer, Information Models can be described by bringing together multiple TOSCA service templates

13 Thank You Your Feedback is Important!
Access the InterConnect 2016 Conference Attendee Portal to complete your session surveys from your smartphone, laptop or conference kiosk.

14 Open Tech @ IBM – Learn More!

15 OASIS TOSCA Backup Materials

16 What is TOSCA? Who participates?
TOSCA is an important new open cloud standard, that is enabling a unique eco-system, supported by a large and growing number of international industry leaders… Associated Companies TOSCA Version 1.0 Specification (XML) - OASIS Standard (Nov 2013) TOSCA Simple Profile v1.0 Specification (YAML) – OASIS Standard (target March 2016) Government and Corporate Awareness: OASIS: participant companies & orgs., participants spanning 65+ countries International Standards & Research: ETSI NFV liaison, EU FP7, etc. Industry Analysts: Forrester names TOSCA as a top four cloud open standard (Mar 2014) Multi-company Interoperability Demonstrated: EuroCloud 2013, Open Data Center Alliance 2014, OSCON 2015

17 Topology Primarily, TOSCA is mainly used to describe the topology of the deployment view for cloud applications Defining Node templates to describe components in the topology structure Defining Relationship templates to describe connections, dependencies, deployment ordering

18 Lifecycle TOSCA models have a consistent view of state-based lifecycle
have Operations (implementations) that can be sequenced against state of any dependent resources fits into any Management Framework or Access Control System Lifecycle Customization Standardize Resource Lifecycle Standardize Relationship Lifecycle source_resource Type_A my_relationship ConnectsTo target_resource Type_B B Lifecycle.Configure.NFV my_resource_name My_Resource_Type Lifecycle.Standard A create Lifecycle.Standard my_pre_config create configure start stop delete Lifecycle.Configure configure Operations start pre_config_source post_config_source add_source remove_source pre_config_target post_config_target add_target remove_target Operations stop my_pre_delete delete Operations Create new Lifecycles or Augment existing (via subclassing) Parameters and Policies can be supplied to operations to affect resource behavior (state) Workflow - TOSCA is developing workflow to allow handling complex state changes, configurations, etc.

19 Choice of Providers with different Infrastructures
Portability from Standardization TOSCA enables applications flexible movement between different cloud infrastructures TOSCA Portable Cloud Application TOSCA Service Template Storage Compute1 DB Compute2 App Network Scaling Policy Avoids binding to a specific implementation by expressing Application Requirements independently from Cloud Capabilities Standardization enables Portability of services across different Clouds Compose from a robust set of Normative Types for different domains (including IaaS, PaaS, NFV) Declarative model spanning infrastructure, platform and application services Manipulate the orchestration declaratively instead of dealing with disparate cloud APIs (leave to the TOSCA Orchestrator) Multi VIM Support TOSCA Orchestration Choice of Providers with different Infrastructures

20 Applying a TOSCA Scaling Policy to Resources in a Cluster
using Senlin and the Kubernetes “redis” example (Pod, Docker Containers) TOSCA Scaling Policy Definition Policy Target In this case, a Kubernetes Pod A TOSCA resource type (tosca.groups.placement type) Normative Properties TOSCA provides base properties used for scaling my_scaling_policy: type: tosca.policies.scaling properties: # normative TOSCA properties for scaling min_instances: 1 max_instances: 10 default_instances: 3 increment: 1 # target the policy at the “Pod” targets: [redis-master-pod] triggers: # Symbolic trigger name resize_compute: event: tosca.events.resource.utilization target_filter: node: master-container requirement: host capability: Container condition: utilization greater_than 80% action: # map to SENLIN::ACTION::RESIZE operation resize: # operation name inputs: strategy: BEST_EFFORT # Invoke Senlin webhook (service) Implementation: Senlin.webhook() TOSCA normative event type (name) that would map to domain-specific names (e.g., OpenStack Ceilometer) Trigger Name Could be used to reference an externalized trigger Map Event, Condition to target Monitoring Service Map event to OpenStack notifications for resource (e.g., Nova) Register alarm with the target monitoring service (e.g., Ceilometer, Monasca, etc.) Monitor Resource Find Resource (node) named “master-container” in TOSCA topology. Locate its “host” Container Attach an alarm | alert | event to identified Container resource and monitor for Condition Parameters optionally provide input parameters, for example “strategy” with value of “BEST_EFFORT” Action Invoke “resize” operation Within Senlin “webhook” (service) Based upon the Senlin “scaling_out_policy_ceilometer.yaml” 20 20


Download ppt "TOSCA Orchestration and Management in OpenStack"

Similar presentations


Ads by Google