Ike A to LF Edge Akraino Comp Stor Netw & Commun & ETSI MEC AppD Mngmnt Present Dec 8th 2022 Rev PA06.pdf

Challenges in (selected) Cloud-native adoption in
(distributed & disaggregated)
MNOs 5G and B5G Networks need
for
Synergy between
Compute, Storage, Networking
and
Communications
Ike Alisson
2022 - 12- 08 Rev PA06

Table of Contents (ToC)
1. Cloud & Communication Issues & Challenges
2. ETSI MEC design & implementation example
3. 3GPP "5G Advanced" (Rel. 18 & 19) proposed Capabilities (evolvement
with e.g. MNOs/ CSPs MOCN "direct" & "indirect" & "equivalent"/inter-
operable NPNs/SNPNs).
2

3
1. Cloud & Communications Systems' (current) Challenges & Issues
Today’s Cloud and Communications Systems are NOT CAPABLE of
- Capturing,
- Transmitting,
- Storing, and
- Analysing
the Petabytes of Data generated by the soon-to-be trillions of Sensors operating 24/7.
They are also NOT PREPARED to deliver the Compute needed for Real-Time AI/ML Inferencing required to drive such
demands that we anticipate will come from our
- FoF (Factory of the Future)
- VR/XR/MR (Virtual, Extended, Mixed Reality and Extended Reality) with Haptic Interactions,
- (V2X) Connected Vehicles,
- Assisted living, or
- Merging of Physical & Digital worlds with 5G & B5G
Ref.: 5GA, DCC, Nov., 2022: 4

4
The Cloud is "Changing"
1st - Applications want to be deployed anywhere & change deployment anytime.
The focus moves from "sharing resources to Composing Dynamic Capabilities,
even after Deployment.
Applications will be Delay- and Latency sensitive, on varying Time-scales
with different Hard- & Soft boundaries.
Communication, Compute, and Storage must be considered as an Integrated Set of Changeable Configurations that
provide the required service to an application.
2nd - “Center of Gravity is moving toward the Devices ("Edge"), & interactions in a Cyber-Physical World best
suited for these tasks and configure any required communication between all end points in important areas such as
- IoT,
- Industry 4.0,
- Private 5G, or
- Retail and Public Services.

5
Management of Resources and Workloads:
Most Orchestration Frameworks today use a Centralized Approach (where)
One (1) Entity has knowledge of all the Resources in the System and Plan how the
Workloads will be mapped.
With the start of Docker & containers, the Kubernetes Project was started to
provide a lightweight & scalable Orchestration solution.
Most existing Compute Systems today, including Edge Computing Systems, rely on
static provisioning.
Thus, the SW & the Services needed to perform the Compute are already residing
at the Edge Server prior to an Edge node requests a Service & the pool of HW
resources is also known a priori to Kubernetes.
This architecture works well for Cloud & the MEC where a Centralized
Orchestration is used.
Since the Resources of the Pervasive Edge are independently owned, the
Orchestration Frameworks need to be extended to handle Dynamic and Multi-
Tenant Resources in a secure manner.

6
Management of Resources and Workloads:
it is important for the Orchestration Architecture to be able to support
Dynamic Discovery & Use of (HW) Resources distributed in the edge.
Kubernetes and Docker are both centralized Architectures, which need
messages exchange and synchronization before a new Service can be
configured on a server.
Hence, new approaches have to be investigated to discover and deploy
new (HW) Resources in Real-Time within the Multi-site & Multi-Edge
Infrastructure of 5G & B5G Systems.
Content Sharing and Resource/Service Orchestration in 5G & B5G
New innovations in terms of Data Movement & the Orchestration of Resource
and
Compute Services will be required.
a few new exemplary approaches on Network - & Application - Layers are
detailed such as …………..

7
2. ETSI MEC design & implementation example
3GPP EAS and ETSI MEC Application Profile Provisioning
The MEC Application can start "producing" or "consuming"
MEC Services after the MEC Application is instantiated &
running.
The Application Information (AppInfo), which can be
regarded as the MEC Application Profile, represents the
information provided by the MEC application instance as part
of the "Application Registration request" message.
Some fields in AppInfo are intentionally not duplicating the
EAS profile (if present) with conflicting parameters but should
be consistent with them.
It can be seen that unlike AppD, which is
mainly used in the Management Plane for
instantiating an Application, and is static in
nature, AppInfo carries the runtime information about the
MEC application instance.
In 3GPP EDGEAPP, the EAS profile is provided in the EAS
registration request. Ref.: 3GPP, Rel. 18 (5G Adv), Nov., 2022

8
3GPP EAS and ETSI MEC Application Profile Provisioning
… in current ETSI MEC specification, no APIs
for MEC Application Registration is defined
because it is assumed that all MEC Applications
are on-boarded and managed by MEC
Orchestrator.
API for MEC Application discovery is not
defined since the existing MEC Service is either
defined from the MEC Application's perspective
or it is consumed by the MEC Application
rather than the UE.
Therefore, the comparison EAS Registration and EAS discovery of
EDGEAPP and ETSI MEC specification shows that:
[Observation 2] ETSI MEC platform(MEP) supports service registration. In the registration parameter
"ServiceInfo"，there is a mandatory field "consumedLocalOnly" used to indicate that the service can
only be consumed by the MEC applications located in the same locality, which means ETSI MEC
services (produced by Authorized MEC APPs) registered and exposed on MEP can be invoked by
MEC consumer APPs deployed on the same or another MEC host.
Ref.: 3GPP, Rel. 18 (5G Adv), Nov., 2022

11
ETSI GS MEC 010-2 V2.2.1 (2022-02)

16
ETSI MEC deployment/commissioning issues/challenges by May-June 2021

22
For the special MEC/Hybrid Cloud Assurance UC - CI/CD/CT,
MEC assurance becomes essential for critical Edge Compute
Applications & Performance & particularly in a Multi-Cloud
Environment at the Carrier/Hyperscale Gateway.
An MEC Validation Platform provides Full Stack MEC Testing &
Performance coverage including Global Security assessment.
This is divided in 3 Main Categories starting from the ground floor
with Cloud Infrastructure Validation including:
1. Capacity & Performance for Latency, Bandwidth & Resiliency,
Benchmarking, Scaling and Secure Access Service Edge (SASE).
2. MEC Nodes Validation need to be conducted for QoS / QoE
Validation, Jitter Latency, Video & Audio Processing, O-RAN
RIC, 5G Core UPF split / N9 interface, xHaul Transport as a
Service, Extended Visibility and Security Assurance
Specification (SCAS).
3. MEC Vertical Services & Applications with QoS / QoE Validation,
Jitter Latency, O-Cloud, Video & Audio Processing, all the
Verticals like C-V2X, industry 4.0, Video surveillance etc. and
Network Security.

23
The O-RAN WG4 Conformance Test Specification
ensures the O-RU’s compliance with the O-RAN
Fronthaul (FH) Standards.
The 3GPP (Test) Specifications requires a full gNB since
3GPP does not recognize the open nature of O-RAN.
3GPP does not separate the Radio from the BaseBand
processing Unit (BBU) as required by O-RAN.
However, it is possible to leverage the 3GPP Transmitter &
Receiver (TRU) Tests (Chptrs 6 & 7 of 3GPP
Specifications) when validating the O-RAN FH.
All test waveforms specified by the O-RAN Conformance
Test Specification use the same test waveforms used in
3GPP tests.
The Test set-up can test a Radio for 3GPP TRU performance and O-RAN conformance. The only difference is that 3GPP
expects the tests to run on a gNB that is in test mode.
The O-RAN tests the Radio using an O-DU emulator and does not require a test mode.
It is not possible to perform 3GPP Chapter 8 conformance tests using the O-DU emulator because it requires MAC layer
processing, which is not present in the O-DU emulator.

24
Autonomous Service Management (enabling
Autonomous Networks) tries to put Analysis &
Decisions into Machines, so it becomes a “Zero-
touch” System for the Operator – more correct: the
operator’s touch moves from the Network to the
Design of the Automation.
• The Management Systems need to provide
Capabilities to enable Autonomy, but also to
accelerate and simplify the DevOps process
• System interactions need to be simplified at API level
and allow autonomous decisions in requested
systems at different levels On the path to full
cloudification, hybrid situations will appear.

27
Ref.: 3GPP 5GAdv, Nov., 2022:

1. List of Equivalent PLMNs, as specified in TS 5G System Architecture
Rel. 17, clause 5.18.2a.
2. The Solution re-use existing Function as specified in clause 5.18.1 of TS 5G
System Architecture, Rel. 17, where different combination of PLMN ID and NID
can point to the same 5GC.
5.18 Network Sharing
A Network Sharing Architecture shall allow Multiple Participating Operators to share resources of a
Single Shared Network according to agreed allocation schemes. The shared network includes a radio
access network. The shared resources include radio resources.
The shared network operator allocates shared resources to the participating operators based on their
planned and current needs and according to service level agreements.
In this Release of the specification, only the 5G Multi-Operator Core Network (5G MOCN) network
sharing architecture, in which only the RAN is shared in 5G System, is supported. 5G MOCN for 5G
System, including UE, RAN and AMF, shall support operators' ability to use more than one PLMN
ID (i.e. with same or different country code (MCC) some of which is specified in TS NAS for UE in
Idle mode and different network codes (MNC)) or combinations of PLMN ID and NID. 5G MOCN
supports NG-RAN Sharing with or without multiple Cell Identity broadcast as described in TS NG
RAN.
5G MOCN also supports the following sharing scenarios involving non-public networks, i.e.NG-
RAN can be shared by any combination of PLMNs, PNI-NPNs (with CAG), and SNPNs (each
identified by PLMN ID and NID).
NOTE 1: PNI-NPNs (without CAG) are not explicitly listed above as it does not require additional NG-RAN
sharing functionality compared to sharing by one or multiple PLMNs.
In all Non-Public Network sharing scenarios, each Cell Identity … is associated with one (1) of the following
Configuration options:
- one or multiple SNPNs;
- one or multiple PNI-NPNs (with CAG); or
- one or multiple PLMNs only.
NOTE 2: This allows the assignment of Multiple Cell Identities to a Cell and also allows the cell identities to be
independently assigned, i.e. without need for coordination, by the network sharing partners, between PLMNs and/or non-
public networks.
NOTE 3: Different PLMN IDs (or combinations of PLMN ID and NID) can also point to the same 5GC. When
same 5GC supports multiple SNPNs (identified by PLMN ID and NID), then they are not used as equivalent SNPNs for a
UE.
NOTE 4: There is no standardized mechanism to avoid paging collisions if the same 5G-S-TMSI is allocated to
different UEs by different PLMNs or SNPNs of the shared network, as the risk of paging collision is assumed to be very
low. If such risk is to be eliminated then PLMNs and SNPNs of the shared network needs to coordinate the value space
of the 5G-S-TMSI to differentiate the PLMNs and SNPNs of the shared network.

29
5G Architecture for Hybrid and Multi-Cloud Environments
The Main Challenges to overcome in a Hybrid & Multi-Cloud Strategy are:
1. Maintaining Portability; 2. Controlling the Total Cost of Ownership (TCO); 3. Optimizing Productivity & Time to Market (TTM).
DevOps – a Set of Practices that brings together SW Development & IT operations with the Goal of Shortening the Development & Delivery Cycle & increasing SW Quality - is
often thought of and discussed in the Context of a Single Company or Organization. The Company usually Develops the SW, Operates it & Provides it as a Service to
Customers, according to the SW-as-a-Service (SaaS) Model. Within this context, it is easier to have Full Control over the Entire Flow, including Full Knowledge of the
Target Deployment Environment.
In the Telecom Space, by contrast, we typically follow the "as-a-Product (aaP) Business model, in which SW is developed by Network SW Vendors e.g. as Ericsson
(Nokia, Huawei, ZTE) & provided to Communication Service Providers (CSPs) that Deploy & Operate it within their Network. This Business Model requires the consideration
of additional aspects.
The most important contrasts between the Standard DevOps SaaS Model & the Telecom aaP Model are the Multiplicity of Deployment
Environments & the fact the Network SW Vendor Development Teams cannot know upfront exactly what the Target Environment looks like.
Although a SaaS Company is likely to Deploy & Manage its SW on two (2) or more different Cloud Environments, this is inevitable within Telco, as each CSP creates &/or
selects its own Cloud infrastructure (Fig. 1 below).
Figure 1: The DevOps and (Telecom) aaP Business Models

Ike A to LF Edge Akraino Comp Stor Netw & Commun & ETSI MEC AppD Mngmnt Present Dec 8th 2022 Rev PA06.pdf

More Related Content

Similar to Ike A to LF Edge Akraino Comp Stor Netw & Commun & ETSI MEC AppD Mngmnt Present Dec 8th 2022 Rev PA06.pdf

More from Ike Alisson

Recently uploaded

Ike A to LF Edge Akraino Comp Stor Netw & Commun & ETSI MEC AppD Mngmnt Present Dec 8th 2022 Rev PA06.pdf