Cloud Native Introduction Balaji Ethirajulu Ericsson Internal | 2018-02-21 5G is use case driven Massive MTC Critical MTC Enhanced MBB Fixed Wireless Access Ericsson Internal | 2018-02-21 Mobile data trafficoutlook Global mobile traffic outlook (EB per month) 160 In 2024, 25 percent of mobile 140 data traffic will be 120 carried by 5G 100 networks. 80 60 40 20 x5 Total mobile data 0 traffic is expected to 2014 2016 2018 2020 2022 2024 increase by five times Ericsson Internal | 2018-02-21 5G – Cornerstones and role of open source — Cloud Native / Microservices — Automation — Orchestration — AI and AI driven Policy Ericsson Internal | 2018-02-21 Ericsson Demo’s – Booth 401 — Enabling Edge Cloud. — 5G Automation & Network Slicing — Cloud native 5G Core & Exposure — Cognitive OSS & AI/ML Platform Ericsson Internal | 2018-02-21 Ericsson – Tutorial Highlights — Cloud native Introduction – Balaji Ethirajulu — Cloud native design principles / Architecture – Tamas Zsiros — A Cloud Native Application Design, Opportunities & Challenges – Henrik Saavedra Persson — Cloud native “Dual-mode 5G cloud core” and Network Exposure (Programmability) – Jitendra Manocha — Cloud Container distribution /Distributed cloud - Sampathkumar Santhanakrishnan — Demo – Show the configuration and installation of CCD/k8’S and also show a scaling on a Kubernetes cluster - Sampathkumar Santhanakrishnan and Russ Mahon Ericsson Internal | 2018-02-21 ONS2019 Cloud Native Tutorial Principles & Architecture Tamas Zsiros Ericsson Internal | 2018-02-21 Cloud native values 5G relies on an evolving ecosystem Evolved Evolved network connectivity, and architecture & application architecture architecture Increased automation & New SW packaging, orchestration support, Devices/Local NW Access Sites Distributed Sites National Sites Global Sites and SW LCM, new evolved hardware, assets to expose leads to shift in Application Cloud functionality between virtual infra and physical infra Management and monetization Network Slices Access, Mobility, Network Applications Cloud Infrastructure Transport Ericsson Internal | 2018-02-21 Cloud native values Speed Scale Efficient Performance operations and capacity ̶ Fast, low-cost ̶ Scale fast at low cost ̶ Automation, no-touch ̶ Optimized capacity introduction of from hundreds of users operations throughput and new services in to millions ̶ Legacy and resource utilization small scale new services ̶ Life-cycle independence (services/infrastructure) Software development in fast-moving, independent and empowered teams Ericsson Internal | 2018-02-21 Design principles Cloud native design principles State optimized Orchestration and Agnosticity Decomposed SW Application resiliency design automation A cloud native application Decomposition enables Change of metrics with Handling state tailored for Reduce or remove the need has to be agnostic to unified and flexible LCM regards to resilliency»: application context and for human intervention to underlying infrastructure » and better resource — Pets vs Cattle state/data type » decrease OPEX and increase speed»: — IaaS utilization » — HA achieved through — Separation of — Bare metal — Microservice distribution and fault application logic and — Internal scaling and architecture state load distribution — Kubernetes based isolation hadleld by CaaS CaaS — Independant lifecycle — Proper state storage management per to guarantee — Network level microservice performance automation through model driven management Ericsson Internal | 2018-02-21 Cloud native application stacks Cloud native Cloud native network App A network App B App business logic App business logic Cloud native Cloud native network App C network App D App business logic App business logic Kubernetes based CaaS: Any IaaS*: Any Kubernetes based CaaS: Any Hardware: Any x86 Hardware: Any x86* CaaS on IaaS CaaS on bare metal — CaaS LCM: similar to any IAAS workload — CaaS LCM: specialized tools (e.g. MaaS) — Virtualization overhead (KVM, VMWare) — No virtualization overhead — Security: harder to get from guest OS to — Security: easier to get from host user space host kernel to host kernel space — Advanced container runtimes with improved separation, i.e. katacontainers Ericsson Internal | 2018-02-21 Cloud Native Application using Kubernetes CaaS — Clusters of Nodes, Pods, Services — The dynamicity applies on many layers Cloud Native Application — Containers, PODs “come and go” Applications Entry/exit and — Higher layers depend on the state of load balancing clusters on lower layers (downward K8S Service K8S Service Services dependency) POD is subject for LCM and scaling — Ideally, every layer should handle its P P POD POD POD downward dependency on its own without PODs O O affecting layers above (e.g. the application D D layer should not care whether the HW is K8S K8S K8S K8S K8S K8S K8S operational or not etc.) K8S Nodes Master Master Master Node Node Node Node — Kubernetes keeps track of (dynamic) cluster members and does general housekeeping VMs or physical around them (starting, stopping, upgrading, machines health checking, etc…) Ericsson Internal | 2018-02-21 PCC and PCG -Micro-service Architectures Packet Core ”Data Model” PCC PCG Pluggable service functions Controller IP address Registration allocation Session TDF and mobility SGW Controller Signaling session Business Logic Business termination Charging I/O support IP session Cache-optimized Logging Configuration Performance Fault load balancing management management management Database O&M NBI/ONAP Generic Services Generic Container as a Service (CaaS) Ericsson Internal | 2018-02-21 Deployment configurations of cloud native applications VMs vs Containers in Telco space? Or rather: IAAS vs CAAS Technologies emerging: Kata Containers, FireCracker, gVisor… Workload isolation o (+ virtlet, KubeVirt…) Compute performance I/O performance o / (PCI-PT / SR-IOV) / (Bare Metal / SR-IOV) LCM operation efficiency o The Kubernetes ecosystem is evolving rapidly and ETSI and Platform maturity / ecosystem ONAP are adopting it VNF package & deployment size Support for (using Kubernetes) cloud native design o VMs & IaaS Containers & CaaS Ericsson Internal | 2018-02-21 Multi VIM -Carrier supported CaaS Management & Orchestration NFVO VNF A VNF B Cloud native Cloud native App business logic App business logic VNF C VNF D App business logic App business logic Guest OS Guest OS c c c c c c c c c c c c c c c c c c c c VM VM VM VM VM VM VM VM VIM K8S IaaS CaaS Hardware Hardware Ericsson Internal | 2018-02-21 Carrier supported CaaS deployed on top of an IaaS Management & Orchestration NFVO VNF VNF Cloud native A B VNF C App business logic App business logic App business logic Guest OS Guest OS c c c c c c K8S CaaS VM VM VM VM VM VM VM VM VM VM VM VIM IaaS Hardware Note: no direct mapping between VMs and containers Ericsson Internal | 2018-02-21 Isolation domains: CaaS deployed on top of an IaaS Management & Orchestration NFVO Isolation Isolation domain domain ”X” Cloud native Cloud native Cloud native ”Y” VNF A VNF B VNF C No Isolation App business logic App business logic App business logic c c c c c c c c c c c c c c c c c c K8S K8S CaaS CaaS Kernel isolation Kernel isolation Kernel VM VM VM VM isolation Kernel VM VM VM VM VM VM VM VIM IaaS (incl. Host OS) Hardware Ericsson Internal | 2018-02-21 CaaS evolution –supporting container and hypervisor workloads Management & Orchestration NFVO Cloud native Cloud native network App A network App B Cloud native Cloud native App business logic App business logic network App C network App D App business logic App business logic Guest OS Guest OS VM VM VM VM VM VM VM VM c c c c c c c c c c c c c c c c c c c c K8S CaaS VM VM VM VM VM VM VM VM VM VM Hardware Improved separation via alternative KubeVirt gVisor container runtimes Ericsson Internal | 2018-02-21 VM VIRTLETS Cloud Native Applications in relation to NFV Service Orchestration Service Orchestration Resource Orchestration & NFVO DC Networking Automation Network Service NS NS EMS VNF VNF VNF VNF VNF VNF VNFM VNF Lifecycle Management Container Infrastructure CaaS CaaS Mgr Management IaaS IaaS Mgr Virtual Infrastructure Management Infrastructure Infra Mgr Infrastructure Management Cloud 2 Cloud 1 Ericsson Internal | 2018-02-21 Orchestration Key building blocks to deliver Cloud Native values Cloud Native WoW and operations Continuous Software Software Orchestration & Automation release distribution upgrade for K8s and Cloud Native NF´s 1 2 3 Acceptance Cloud Native NF´s Ericsson Customer 4 tests 6 5 K8S based platform Data Feedback collection Any Infrastructure Ericsson Internal | 2018-02-21 Conclusions — 5G use cases rely on flexibility of cloud native — Cloud native transformation is a holistic chalenge — Application architecture — Cloud Infrastructure — WoW & Operations — Cloud native design principles — Agnosticity, disaggregation, resiliency, state optimized, automation — NFV ecosystem compatibility is key Ericsson Internal | 2018-02-21 Cloud Native Application Design Opportunities & Challenges Henrik Saavedra Persson Ericsson Internal | 2018-02-21 Ericsson application development framework Application development framework Application specific services Reference Container platform, 2 1 K8s based Generic Services Mgmt. Application Data Security Network Mgmt Monitoring Management & Orchestration stack & Service Services Services Services Services Services 3 Mgmt. 2 functionality for Services and Applications 4 Application development & onboarding environment Common (platform 3 type/generic) services for reuse