This page captures the details of the Modeling enhancements (IM, DM of NSST, NSSI, service and slice profiles, etc.) and the associated component impacts (e.g., SDC, AAI), as a part of E2E Network Slicing.

Details of SDC Template design for Option 1:  Template Design for Option 1

Details of SDC Template design for Option 2:  Template Design for Option2

Details of A&AI Instance model: A&AI Instance Modeling in Guilin

Main Contact(s): Chuyi GuoBorislav Glozman

Participants: LIN MENGSwaminathan SeetharamanSeshu Kumar MudigantiChuanyu ChenShankaranarayanan Puzhavakath Narayanan,Fei Zhang Alessandro Gerardo D'Alessandro


Enhancement Aspects

E2E Model structure optimization

  1. New/Enhanced features
    ( Blue means new in Guilin. )

    • Introduce Slice Profile Templates to carry user slicing subnet requirements, in ONAP we have 3 kinds of Slice Profile Templates, for RAN, Core and CN separatedly.
    • Add RAN NSST and TN NSST to support RAN and TN NSSI deployment, in TN side, also add TN Network Requirement Template to support TN connection links creation.
    • Has allotted resource relationship between slice subnet requirements and subnet slice resource, see 1, 2, 3. Allotted resource 4 is inner TN, between subnet resource and logical network resource.
    • Update relationship between NST and NSST using allotted resource, instead of nested, see 5. 
  2. Supporting for both option 1 and option 2
    ONAP option 1 and 2 differs on the RAN and FH/BH relationship, in option 1, RAN NSSMF will decide FH/MH requirements and send to TN side, so RAN NSST will nest FH/MH NSST, and slice profile requirements will contain not only RAN side requirements but also FH/MH. In option 2, FH and MH are independent requirments when doing E2E requirements decomposition. Detailed background of option 1 and 2, please see  E2E Network Slicing Use Case in R7 Guilin.

3 domain combination solution -Endpoint 

  In order to enable 3 subnet connectivity, making end-to-end forwarding and control plane interacting, in slicing services, we need slicing interworking identifier and information to tell how RAN and Core connect with TN. Slicing interworking identifier(in ONAP is vlan Id, allocated by NSMF) is used to isolate differerent service datas, and identify the unique slicing service across RAN, TN and Core, 1 S-NSSAI will need to be assigned with 1 slicing interworking identifier(vlan Id).  The slicing interworking  information is used as bridges to connect RAN/Core and TN, it should be RAN/Core side outgress and TN ingress, like port, ipAddress, next hops and etc.

  In ONAP, we refer 3GPP proposal 3GPP S5-203419 and S5-203420.rar in TS 28.541, and introduce Endpoint(EP) to ONAP Network Slicing model.

3GPP reference:

ONAP Model:

Design Time: In R7, EP allocation starts from UUI, when doing service profile decomposition to slice profiles, NSMF will give EP information along with slice subnet requirements delivery to RAN and Core, TN will get the information and store EP Ids after RAN and Core instantiation.

Run Time: 

Run time model reuses node NetworkRoute to carry EP information, 1 NSSI can have multiple EPs, 1 EP will have a Slicing interworking identifier(vlan Id).

(Note: 1 EP can support multiple S-NSSAIs, but in R7, we only support EP and S-NSSAI 1 : 1.)

EP attributes and AAI NetworkRoute schema mapping:

Endpoint(EP)NetworkRoute TypeDescription
idroute-id

String

Identifier of this EP.

ipAddress

ip-address

String

This parameter specifies the IP address(IPv4 address or IPv6 address) assigned to this endpoint, it can be DC gateway IP address.

logicInterfaceId

logic-interface-id

String

This is used for Slicing interworking identifier, it could be VLAN ID, MPLS Tag or Segment ID.

nextHopInfo

Next-hop

String

This parameter is used to identify ingress transport nodes identification. This can be any of combination of IP address of next-hop router of transport network, system name, port name, IP management address of transport nodes.

Add RAN Model


Add TN model

Transport Network Slicing refers IETF TSCi model, in ONAP design time, there has TN NSST and TN Req T, in run time, reusing logicalLink netwokPolicy and etc. to express IETF nodes, details please see  

Supporting Files

Model Design Files

Description

File

Changes
WIP model of RAN part
Design Time model-Update Core model-(This sildes include latest RAN part)Fixed a typo
WIP Modeling files (Aug 6)
WIP Modeling files (Aug 14)
  1. Change EP design: NSST will have EP information while slice profile not.
  2. Change attribute(serviceRole and etc.) values.
  3. Add all RAN side templates(4 in total). 
WIP Modeling files (Aug 20)

1.DLThptPerSlice, uLThptPerSlice, maxNumberofConns and maxNumberofPDUSession

2.EP for supporting FH and MH

Completed set of files for both option 1 and 2 (Nov 16) 

Remove Endpoint from NSST,  Endpoint information will be in slice profile Ts.


Design Time Template


Note: in Guilin release, option 2 only supports BH TN.


Description

File

Changes
Guilin E2E Network Slicing Service Templates (Aug 28)
WIP RAN Option 1 (Oct 4)


AAI Instance Model

Description

File

Changes

AAI model proposal.

Including a E2E design and 3 domain subnets instance model.


AAI model proposal update
  1. Update different domain view(also separate option 1 and 2)
  2.  Update the attributes mapping for idenifying option1/option2, external/internal.

Final attributes mapping for service instance

update the latest changes of AAI, reusing Network-route to carry EP information, adding modeling overview picture 

Service Profile and Slice Profile Attributes Mapping in R7

Slice Profile

Attribute Name Derived

From Service Profile

DomainValueSupport

sNSSAI

sNSSAI

All

Copy

M

pLMNIdList

pLMNIdList

All

Copy

M

maxNumberofUEs

maxNumberofUEs

CN AN

Copy

M

coverageAreaTAList

coverageArea

AN

Transfers into TA ids

O

latency

latency

All

Derived from

M

uEMobilityLevel

uEMobilityLevel

CN AN

Copy

O

resourceSharingLevel

resourceSharingLevel

CN AN

Copy

M

sST

sST

All

Copy

M

maxBandwidth

dLThptPerSlice,

uLThptPerSlice

TN

Derived from

O

expDataRateDL

(eMBB)

dLThptPerSlice

CN AN

Copy

O

expDataRateUL

(eMBB)

uLThptPerSlic

CN AN

Copy

O

areaTrafficCapDL

(eMBB)


CN AN

Calculate internally

O

areaTrafficCapUL

(eMBB)


CN AN

Calculate internally

O

maxNumberofPDUSession

maxNumberofConns

CN AN

Copy

O

overallUserDensity

(eMBB)

termDensity

AN

Copy

O

activityFactor

(eMBB)

activityFactor

CN AN

Copy

O

jitter

jitter

TN

Derived from

O

survivalTime

(uRLLC)

survivalTime

CN AN

Copy

O

transferIntervalTarget

(uRLLC)

No mapping in R7


CN AN


O

cSAvailabilityTarget

(uRLLC)

availability

CN AN

Copy

O

cSReliabilityMeanTime

(uRLLC)

Reliability

CN AN

Copy

O

expDataRate

(uRLLC)

dLThptPerSlice,

uLThptPerSlice

CN AN

Derived from

O

msgSizeByte

(uRLLC)

maxPktSize

CN

Copy

O

Note: Support "M" means this attribute must have a value in the slice profile when doing decomposition, "O" means may have a value or not depending on customer requirements.



  • No labels

2 Comments

  1. Hi Chuyi GuoBorislav Glozman,

    I am referring to slide AAI proposal on E2E Network Slicing for R7-V1.4 - 0721.ppt

    For core NSSMF implementation we have a requirement to store both the design time ID (model invariant id) and service instance ID (after instantiation) of the Network Service.

    As per slide #7, it seems NSSI shall have reference to NS service instance id. However not able to locate the actual field that shall store this. In slide #12 the NSSI has two new fields service-function and workload-content, but not sure of it's intent. 

    Also there seems to be no provision to store the NS design time ID (model invariant id) during the slice design phase. The SliceProfile may have to be modified to include the NS model invariant ID to be stored.


    Please do let know your thoughts on the above.


    Thanks.

    1. Hi Milind,

        My understanding is NSSI can know the related Network Service instance through relationshiplist, when you check NSSI, you can find the associated Network Service instance.

        At Network Service side, as you said there will store both the design time ID (model invariant id) and service instance ID (after instantiation) in Network Service instance.