Releases: anuket-project/anuket-specifications
Releases · anuket-project/anuket-specifications
Orinoco
| Sub-project | Contact | What has changed |
|---|---|---|
| RA2 | Riccardo Gasparetto Stori | GSMA Publication - NG.139 docx created and submitted to GSMA, Update Kubernetes version - to 1.26, Add Express Data Path (XDP), Fixes: clarify gaps, security requirements |
| RM | Walter Kozlowski | Added section on O-RAN alignment, O-Cloud requirements, Issue #3297, Added information about Security HW assist, Issue #3296, Introduced AF_XDP technology commentary, Issue #3314, Added Hugepages metrics to Table 4.8, Issue #3224 |
Nile
Nile
Moselle
| Workstream | Contact | What has changed? | Significance |
|---|---|---|---|
| Anuket Kubernetes Reference Architecture (RA2) | Riccardo Gasparetto Stori | Added CNF workload guidelines, aligned with CNCF CNF certification test listadded CNF packaging specificationsAdded Edge Cloud Hardware requirementsUpdated to Kubernetes 1.23Added Host OS requirements | Enables CNF providers to follow industry guidelines to ensure containerised workloads follow cloud native best practices and can run on Anuket compliant cloud infrastructure. These CNFs can be then easily deployed by operators.Standardised deployment practices include packaging.Relaxes strict requirements on hardware performance allowing constrained edge deployments - important for 5G.Kubernetes 1.23, the latest supported release of Kubernetes, includes important new features relevant to telcos such as IPv4/IPv6 Dual-stack, HorizontalPodAutoscaler v2 and Generic Ephemeral Volumes. Support is also extended until Dec. 22.Host OS requirement allow cloud providers to identify the minimum OS & kernel capabilities required to run Anuket compliant Kubernetes platforms, and leverage the latest security and other improvements. |
| Anuket Reference Model (RM) | Walter Kozlowski | Multi-cloud interaction summary table and examples section "Requirements, Reference Architecture & Industry Standards Intersect" addedSingle pane requirements added (for Multi-Cloud)Software supply chain security updatedRegulatory directives for cybersecurity addedStorage: Tenant Reference, storage types, use cases and stereotypes updatedInfrastructure LCM Automation section addedGeneral PaaS Services contents updated (Logging, Monitoring, LB requirements)Specs for infrastructure monitoring from ETSI GS NFV-TST 008 V3.5.1 added | Gives Communication Service Providers (CSPs) a base to adopt and sponsor a set of standards that are necessary to support the interactions with Cloud Providers.Provides CSPs with a minimum set of requirements for operations in a Hybrid, Edge, and Multi-Cloud environment.Software supply chain security is crucial and is made complex by the greater attack surface provided by the many different supply chains in virtualised, containerised, and edge environments. This section will help the operators to implement the best practices to address these risks.Provides leading examples of regulatory frameworks relevant for Cloud Infrastructure in the telecom environment. This new section will provide important guidance for the operators.Provides a relatively simple way for the storage consumer to specify / select their storage needs.Provides operators with a structured framework and best practice principles for Automation of Infrastructure Life-cycle.Provides a telecom operator (CSPs) with a set of new or updated requirements for selected non-Telco PaaS Services.Fleshes out the specs for infrastructure monitoring from ETSI GS NFV-TST 008 V3.5.1, to be used by RA-1 and RA-2. |
| OpenStack Reference Architecture (RA1) | Karine Sevilla | Addition or update of specifications reference in tables of requirementsEdge : OpenStack services deployment, storage scenarios, image caching considerationsStorage: developments aligned with Reference ModelLCM automation: installers description, automation of underlying resources provisioning | The reader will easily find details of configurations or specific information with pointers to sections addressing the requirements. It simplifies the use of the specification. Guidance for edge deployments, including relaxation of Inputs for the choice of storage solution per workload requirementsGuidance for automation, examples of tools |
| RI2/Kuberef | Dan Xu | Align requirements covered in RI2 with RA2 Lakelse, and support 16 new requirements in RI2.Did some bug fix and automate some prepare steps in kuberef. | Automated e2e deployment of a reference platform based on RA2 Lakelse on baremetal, infrastructure provider (Equinix Metal) and virtualized infrastructure using opensource tooling. |
Lakelse
So, what specifically have we achieved in Lakelse?
Numerous Anuket projects and work streams continued their efforts to strengthen container-based open infrastructure specifications and implementations – an area that is increasingly important to the telco industry. Some of the efforts include:
- ViNePerf has significantly enhanced its network performance testing framework to support Kubernetes use cases resulting in the Globecom 2021 Conference Paper, “Benchmarking Kubernetes Container-Networking for Telco Usecases”
- Barometer improved automation, enabling easier testing and configuration of the metrics collection
- CIRV added 12 Kubernetes post-deployment validations
- And finally, a new project Thoth to investigate using AI/ML methods for some NFV use cases starting from failure prediction.
On the requirements side
Both RA1 and RA2 follow the Reference Model (RM) efforts in evolving open cloud infrastructure specification to include Hybrid, Multi-Cloud model, with significant updates to interaction models, security, and life cycle requirements. Security for such models has been and will continue to be subject of a very intense discussion in the telco industry. Security starts at the infrastructure level, so there is a growing interest in Anuket outcomes and more collaboration. One effort in this area: GSMA Fraud and Security Group (FASG) work on 5G security references the Anuket Reference Model, and the collaboration with FASG is growing.
Hybrid, Multi-Cloud model and AI/ML utilisation, will be in my view a major theme of Anuket work in the upcoming releases, reflecting a growing global interest in more sophisticated operating models for telecommunication operators facing the challenges of real-life 5G/Edge/IoT implementations.
Kali
Reference Model:
- New flexible, hierarchical model (infrastructure profiles and extensions, workload profiles and flavours) to support evolving technologies and a variety of deployment styles
- Hybrid Multi-Cloud model (Data Centre to Central Office, to Edge), including Telco Edge Model
- Abstraction and normalisation of hardware acceleration
- New security requirements including IaC Security, DevSecOps best practices for Infrastructure, Zero Trust, Open Source Software Security
Reference Arch 1:
- Cloud Topology changes including Distributed control and Edge Content
- Major enhancements to the Security Requirements aligned with RM
- Aligned with the hierarchical profiles introduced in RM
- Changes to address concerns, for example, in the areas of networking
Reference Arch 2:
- Explicit definition of Kubernetes release, and API policy
- Define mandatory features as part of the new Special Interest Group requirements chapter, aiding automatic conformance testing
- Update to K8s 1.21
- Added custom resources and controllersAdded device plugins and accelerators
- Differentiating CaaS-on-VM versus Bare Metal deployments
See https://wiki.anuket.io/display/HOME/Kali+Release+Highlights
Elbrus
Reference Model:
- defined Hardware Infrastructure Management and initiated collaboration with the LFN ODIM project on possible implementations
- initiated discussion on the abstraction and normalization of the hardware acceleration and network fabric programmability
- developed a lot of contents (more to come) on the topical subject of Infrastructure Automation and Observability
- started a new chapter “Hybrid Multi-Cloud model: Data Centre to Edge” to address the evolution from the pure Data Centre based NFVI into a much more realistic (and complicated) deployment models
