Network Service Management and Orchestration

Most IT and network professionals are familiar with the value of automation.  Using tools like Ansible and Terraform, a single technician can complete infrastructure and network deployment more rapidly and efficiently. But such methods are "fire and forget" leaving no database of changes or affected devices.  Service management requires more.  

Service management requires the automation system "remember" every change made and every resource consumed.  To effectively manage services, a network operator must have a database of all the services deployed, the resources associated with each service and a way to modify or "roll back" a service change. Such automation requires a different set of tools, technology and integrations.  

WWT has expertise and experience with such tools and systems.  We can help you understand the technology and tools used to orchestrate multi-vendor, multi-domain services.  We can get you started on your service orchestration journey and carry through to completed deployments.  

For a more in-depth look at Service Orchestration technology, please check out these articles:  

https://www.wwt.com/article/orchestration-strategy-next-level

https://www.wwt.com/article/what-is-model-driven-orchestration-and-why-would-i-use-it

https://www.wwt.com/article/cloudify-i-call-it-ansible-for-cloud

Network operators of all sizes find compelling technical and financial reasons to employ service orchestration and service management. 

• Time To Market for new services: getting new services out to your markets rapidly and accurately
• Reduction of human errors: model-driven architectures ensure consistent service deployment
• Improved inventory management: eliminate stranded ports, unused channels and under-utilized circuits
• Competitive advantage: gain the upper hand against competitors with more advanced back-office systems
• Customer experience: measure, improve and create customer delight with metrics and automated assurance
• Introduction of new technologies: wrap your new technology offerings into your existing fulfillment processes

Market Factors: get new services out to your markets and start monetizing your infrastructure investments.  Service Orchestration can ensure those new services are deployed accurately and are integrated into your existing fulfilment systems.  Introduce  new technologies faster with orchestrated deployment. Create new offerings which include cloud, network and content - all delivered via orchestration and integrated into your ordering and marketing systems.  Gain the upper hand by getting your services to market faster and with greater accuracy and fewer errors. 

Operational Factors: reduce human errors by minimizing "keystroke" operations. Create automated workflows integrated with underlying orchestration tools, thus reducing human actions and ensuring clean, consistent service delivery.  Manage multiple datacenters, networks and all the associated applications from a single dashboard.  Have system control over day 0, 1 and 2 operations across your footprint.  Rapidly deploy multi-domain solutions from a single system, across public and private cloud, network and applications. 

Quality Factors: collect relevant KPI information and then craft automated response to network conditions. Use closed-loop automation to make your network react before customers even know there was a problem. Monitor network traffic flow and adjust traffic patterns to avoid bottlenecks.  Quantify and monitor customer experience to improve retention.  

Here is a case study where a mid-market service provider used service orchestration and WWT services to get to market faster:

 https://www.wwt.com/case-study/midmarket-wireless-service-provider-accelerates-automation-with-itential

Service Orchestration  is typically done with "model-driven" type tools.  Systems like Cisco's Network Service Orchestrator (NSO) are key to service orchestration solutions.  First, let's consider typical "file-driven" tools like Ansible.  With Ansible, you create a file called a "playbook".  The playbook contains a list of instructions which tell the orchestrator to conduct various operations (typically over CLI connections to the target devices).  The playbook executes and the results are reported.  Important to note: the orchestrator (in this case, Ansible) does not retain a database of all of the changes made.  Such a tool is very good for deploying new infrastructure (push button, get network) and for doing certain admin tasks like upgrading devices.  

Now, consider what you do when you deploy a service like a VPN:

• You decide on a list of endpoints in the network
• Configure routes
• Configure QoS parameters for this service (bronze, silver, gold, for instance)
• Open access lists for the customer
• Add the customer to servers and databases for the purchased services
• Add the service to monitoring systems for quality reporting

All of those actions must occur every time the service is deployed, also all those configuration items (CIs) must remain associated with that customer.  If the customer adds to the service (like adding a new drop) those CIs must be added to the existing service record.  If the customer cancels the service, then all of those CIs (and the associated consumed inventory like ports) must be released.

As you can see, this is a much more complex operation than turning up a device.  Service Orchestration tools (like NSO) achieve these operations through the use of a "service model".  The service model contains all of the intelligence needed to deploy the service.  The operator is asked to provide variables like "what are your endpoints" and "what QoS profile to apply", then the orchestrator calculates all the CIs for each device. The orchestrator tests the deployment against its database (a dry-run) then deploys the CIs to all the target device in a single transaction.  If anything goes wrong (a device is unreachable, perhaps) the entire transaction is "rolled back" and no network resources are consumed. If the transaction succeeds, the service is marked in the orchestrator's database.  The service can now be managed through its entire lifecycle: deploy, change, remove.  

Cisco Network Service Orchestrator:  

Cisco's Network Service Orchestrator (NSO) is the premier model-driven orchestration solution on the market.  Tier 1, 2 and 3 service providers, Fortune 100 and 500 enterprises, federal and military customers have made NSO the most trusted solution in the model-driven orchestration space.  

WWT can help you get started with NSO with NSO training

Get hands-on with NSO using our NSO labs:

https://www.wwt.com/lab/networkserviceorchestrator

https://www.wwt.com/lab/network-service-orchestrator-nso-creating-a-basic-service-model

Learn more about NSO from Cisco:

A demo showing how NSO manages service lifecycle

Itential Automation Platform:

The Itential Automation Platform (IAP) is a leading workflow and automation solution.  Itential has a 10 year track record of providing innovative, agile solutions.  Itential can take those operations where you use multiple tools across multiple workgroups and turn it into a single, controlled process with full metrics.  Itential works with almost any underlying orchestrator (NSO, Ansible, Terraform, your scripts, etc.) and gives you a way to graphically create a workflow specific to your operation.  In addition to this, Itential has loads of pre-built content and adapters available on their web site. This helps speed development and empowers you to build your own solutions and get to market faster with less cost. 

Learn more about Itential:

https://www.wwt.com/article/itential-configuration-management

https://www.wwt.com/case-study/midmarket-wireless-service-provider-accelerates-automation-with-itential

https://www.itential.com/

Cloudify:

Full, turnkey DevOps automation is here!  Cloudify gives you a way to automate datacenter, application and network deployment across day 0,1 and 2 scenarios. Cloudy acts as a "manager of managers" in a single datacenter or multiple datacenters, across private and public cloud and network infrastructure. Use cloudify to deploy workloads into any Virtual Infrastructure Manager (VIM) like VMware, OpenStack, AWS and Azure.  Along with workloads, deploy networks and applications, then bring the whole deployment under monitoring.  Overlay any day 1 and 2 changes with the same tool over the same deployment.  Then manage the entire deployment globally across multiple datacenters and public cloud infrastructure.  

Learn more about Cloudify:

https://www.wwt.com/article/cloudify-i-call-it-ansible-for-cloud

https://www.wwt.com/article/get-edge-ready-by-deploying-a-next-generation-central-office

https://cloudify.co/

Cisco Crosswork:

Bring the power of automation to your network management and assurance strategy with Cisco Crosswork.  Crosswork is a suite of tools which work together to collect KPI data, visualize and manage that data and create reactive scenarios for true closed-loop assurance automation.  The Crosswork suite includes:

• Cisco Network Controller - intent-based, multi-vendor network and service provisioning 
• Hierarchical Controller -  Extends routed optical networking into multilayer and multivendor environments
• Optimization Engine - Provides real-time network optimization to maximize network utility and improve customer experiences
• Health Insights and Change Automation - Check network health against key performance indicators with automated, closed-loop remediation

Learn more at:

https://www.cisco.com/c/en/us/products/cloud-systems-management/crosswork-network-automation/index.html

A: Services are a dynamic collection of Configuration Items (CIs) across many devices.  A single service may include config lines across endpoints, core, security and datacenter.  All those CIs are considered a single service which has a "lifecycle".  So, you may add to the service, change service parameters or remove the service.  Service orchestrators manage all the CIs as a single entity and allow you to modify, rollback and remove the service cleanly from all devices.  Infrastructure automation rarely has such requirements.  So, this is why there are two classes of tools in the market - infrastructure automation tools like Ansible, Terraform, Puppet, Salt and Chef; and service orchestration tools like Cisco's NSO product, Juniper's Anuta and Cloudify.  Here is an article Model Driven Orchestration which goes deeper into this topic.  

An orchestrated service: in the above diagram, a VPN connects to various endpoints (ce6, ce3, ce4 and ce5).  It also has config in core routers (pe2, p3,p2, pe3 and p0).  There is also various QoS applied (Bronze and Silver) in different parts of the VPN.  The orchestrator (in this case NSO) must keep track of all config lines for this service and manage all CIs across all devices.  The orchestrator allows for change, rollback and removal of the service in an idempotent fashion (only changing what needs changed).  

A: That depends on your solution architecture.  Doing lifecycle management requires that you have a persistent record of each service with meta markers on each CI so you know when that CI was added.  Ansible doesn't have an associated database with meta markers.  So, you could use Ansible to deliver config changes to devices, but it would have to be part of a larger integration with a database and custom code to add and manage meta markers.  

Custom Automation Solution vs. COTS: Here we have a service which has CIs across core and edge devices.  To effectively manage the service, you could integrate Ansible, Ansible Tower and a database using custom developed code.  A COTS product like NSO has all needed functionality to achieve the same result.

A: The truth is it's not an "either or" proposition. There are many cases where opensource infrastructure tools like Ansible are the best choice.  They are simpler to use, have multi-vendor support and no cost to get started.  However, the requirements for service orchestration may drive you to a tool like NSO which is a model driven orchestrator and has multi-vendor support with rich APIs. Here is an article on What Is "Model-Driven" Orchestration -- And Why Would I Use It?

A: Cloud automation solutions do have overlap with one another, but there are differences which may guide your decision.  Many practitioners begin with a purely opensource system like HashiCorp's Terraform.  But they may find that other operational requirements will drive them to COTS (common off the shelf) systems like Cloudify, CloudBolt or Morpheus Managing service in cloud environments can be an overlay process (as it is in networks); or it can be more of a day-1, day-2 process.  Here is an article that discuss the 5 Reasons To Consider a Full Service Cloud Manager.  

A: DevOps (the ability to create code and have it automatically pushed to the operating environment) has revolutionized IT delivery.  Tools like Ansible and GItHub are critical to successful DevOps implementations. 

However, applying these same principals to service bearing networks with clients (who have SLAs) is a very risky proposition. Making network and service changes is a highly controlled process, typically only executed in maintenance windows.  The ability to roll back a change is vital, since a change may cause a costly outage.  

In cloud environments, applying DevOps principals is becoming very broadly adopted.  Here is a demo showing how Cloudify can be used to apply changes to cloud environments based on new blueprints which are posted to GitHub.  ServiceNow has included Cloudify in its integration catalog because Cloudify elegantly supports DevOps 

A: WWT offers hands on training and labs for NSO.  These ATC resources are a great way to get started with model driven orchestration.  They're also a great way to get prepped for taking Cisco courses on NSO. 

• NSO lab (do this one first) - This lab will get you familiar with NSO and model driven orchestration 
• NSO: Creating a Basic Service Model - This lab will walk you through building your first service model 
• We also offer NSO training - This course can be done on-line or in person