Showing results for 
Search instead for 
Did you mean: 
Cisco Employee

Service assurance package in NSO ?

Hi Bilal,


I need a quick favour for Singtel NFVI tender.


A question came about designing telco grade NFVI environment and I need to dream up literally an answer about 5 9’s availability. I wanted to say that it is not all about NFVIoffers and also I thought the service assurance package that NSO has might help.

Would you mind giving me a brief description of how that package works and what its benefits are, so I can include it in the overall response ?


Thanks a lot in advance,


Cisco Employee

Hi Rada,


Copying Stefan to keep me honest here.


We do have a term floating around called “orchestrated assurance”.  It may mean different things to different people. I don’t think there is one single “package” that will work for *all* possible services built for various customer use-cases.


My understanding is as follows:

1.      At service design time, the designer spends some extra time to think about the “metrics” that should be used to measure the “end customer experience” while consuming that service. For some service it could be as simple as doing PINGs for others it could be jitter/delay or something more sophisticated/elaborate than that.

2.      Then the service designer thinks about how these metrics should be collected. Do we do “show commands” on routers/switches. Can we do SNMP queries. Or whether special “virtual probes” need to be instantiated at critical points in the network.

3.      The service designer then models these service-assurance “metrics” from (1) and also the thresholds that should not be breached (or some reactive “action” that needs to be taken when the thresholds are breached).

4.      Service designer develops the assurance “logic” as part of service-package development work. This logic includes setting up the required probes (vProbes or IP-SLA sessions on routers/switches, etc.) and implements “trigger” handlers so that when thresholds are breached and trigger events are received by NSO, the assurance logic can take the reactive actions. We use our patented Reactive FastMap pattern to keep the amount of logic/coding required to a minimum.

5.      Then at service-instance provisioning time, both the service-instance itself and the corresponding assurance is orchestrated. When the service-instance/tenant is decommissioned, NSO will automatically remove all the assurance constructs that were provisioned earlier in the network.


I guess what I’m trying to say is that there is no silver bullet. Rather NSO platform allows for the above assurance portion to be built along with the service-provisioning. Very different from the siloedworked.


Does the above help. Sorry about the long worded explanation, but I thought I give a more realistic picture of what is being offered. Below is a 21 minute recording of Stefan’s WebEx which covers what I have







(cc:ing service orchestration list, useful info)



Thanks for the excellent summary Bilal!



Here is a new video, deep dive on the first one:


(First video here:



Two things to add/emphasise:




One thing to stress on top of Bilals summary below: focus on service KPIs and ways to measure service KPIs directly.


The assurance industry has been too focused on collecting device oriented KPIs and trying to calculate service KPIs as an after-thought.


In the process described by Bilal below, in the service model you have service KPIs specific for your service type.


Like a VPN service has jitter, delay, loss, a video service has buffer size etc.



Ask a network engineer, how do you measure that when trouble-shooting? This seldom leads to SNMP polls.


That is why IP SLA, test probes,  Operational commands in CLIs etc is useful.


But drive this top-down, by the relevant service KPIs you defined in your service model.





NSO should do


a) provisioning, classical


b) serious activation test, let the service model have an activation-test action, intrusive.


    so before handing over the service to the customer, NSO should have run a serious activation test, not just ping


c) turn on assurance, trigger external assurance systems to monitor the KPIs defined by the service model



What is important is that all this is part of Reactive Fastmap so that all changes on the service instance is reflected in the assurance systems



Screen shot from latest demo below:


br Stefan

Cisco Employee


Stefan, Bilal, thanks for this very useful interaction.


At the end of the day what counts is overall business impact. Few SPs are doing a relatively good job in building an hierarchical KPI constructs that aggregates devices specific KPI’s into Service KPIs and Business impact.


Below an example of a project Cisco (former IBSG) has done for a US based Cable operator.





Cisco Employee




Stefan Vallin message is correct indeed we need to add that Cisco is working on a companion platform of the orchestration system with the goal  to provide customer facing service assurance. This platform is going to be based on the following principle that are required for next-gen OSS system:


(1) Open and modular, aligned with Big Data framework
(2) Analytics based OSS functions applied across physical and virtual infrastructure
(3) Service Model driven assurance – i.e. the concept covered by Stefan and by the below slide.


Recognize Your Peers
Content for Community-Ad