Download presentation
Presentation is loading. Please wait.
1
Applications Commiteee
Sept 2017 Version Pascal Menezes CTO
2
The Digital Economy Complete Retooling of Networks for a Digital Economy A New Economy Hyper-connected On-Demand and Agile Assured and Secure Private and Public Clouds Machine Automation
3
Orchestrated Services
MEF’s Vision Orchestrated Services Security Services Optical Carrier Ethernet IP Services L4-L7 Point Services App Services SD-WAN Agile, Assured and Orchestrated Service Provider ”X” Service Provider “Y” Mobile Residential Cloud Provider Data Center Internet Enterprise SDN NFV Legacy Networks Self-service Web Portal Framework Framework
4
MEF 3.0 Global Services Framework Launching Nov – 2017
Orchestrated Services Carrier Ethernet L1/Wavelengths L3 VPNs SD-WAN Security-as-a-Service Application Services Open LSO APIs LSO APIs for orchestration across Multiple service providers Multiple network technology domains Community Reference Implementations Certified Professionals MEF Developer Community Enterprise Advisory Council Hackathons Pro Cert Open Source Projects and SDOs On-Demand Certification Cloud-based test platform Subscription-based certification of services & LSO APIs
5
MEF’s Member Network Transformational Journey
3.0 Orchestrated Multi-Connectivity Services 3.0 Multi-Connectivity Static Services 3.0 Orchestrated Virtualized Services 3.0 App Aware Network Services Static CE 3.0 CE Grow the market to $100B Bring in T2-4 operators Bring CE up to date from 2012 specs to current New virtualized and elastic CE access services New 3.0 CERTaaS model for CE 3.0 Experimenting with LSO and APIs to become orchestrated ready New 3.0 Professional Cert generalist program for training CSP staff on up to date CE 3.0 LSO, SDN and NFV Static IP, Optical and SD-WAN service definitions Manual inter-provider services at the NNI 3.0 CERTaaS multi-connectivity cert SDN 3.0 Pro Cert Generalist program Orchestrated CE, IP, SD-WAN and Optical services Intent LSO APIs Implementation using MEF LSO APIs 3.0 CERTaaS orchestrated multi-connectivity cert LSO 3.0 Pro Cert Specialist program Orchestrated SECaaS and L4-L7 virtualized services Implementation of MEF LSO APIs for virtualized services 3.0 CERTaaS virtualized cert NFV 3.0 Pro Cert Specialist program App assurance services AI closed loop analytics Intent App APIs 3.0 CERTaaS App cert Analytic 3.0 Pro Cert Specialist program Static CE 2.0 App Aware Networks MEF 3.0 CE 3.0 Multi-Connectivity Services Orchestrated 3.0 Multi Connectivity Services Orchestrated Virtualized Services $80B Market Multiple Retail and Wholesale Services Massive success on certification
6
Challenges with the IT Cloud Transformation
Service Provider Network Real-Time SaaS Applications Cloud Network Application Packets Device UC&C WebRTC Enterprise Network Video Surveillance AR/VR Video VTC VoIP Device Mobile Network Connected Cars Smart Homes M2M As IT continues its Cloud Transformation journey, the network connectivity between cloud hosted applications and user endpoints (enterprise, mobile or residential) has challenges with Complexity of the underlying infrastructure – There is a complex ecosystem of moving parts between device and cloud application often times getting manual treatment in terms of configuration & management. Lack of Visibility. - Very complex to troubleshoot when an application does not perform Lack of SLA Guarantees - . Lack of Awareness - Applications and networks lack any awareness of each other Smart Cities Device Residential Network IoT SaaS Applications Lack of SLA guarantees, Lack of Visibility, Highly- complex
7
Service Provider Network
What we need Service Provider Network Real-Time SaaS Applications Cloud Network Application Packets Device UC&C WebRTC Enterprise Network Video Surveillance AR/VR Video VTC VoIP Device Mobile Network Connected Cars Smart Homes M2M Agile: Program networks for autonomic behavior such that - control, visibility, and performance on-demand and in near real-time. Assured: Enforce network delivery of application and/or service performance objectives, availability, security, and middle box interoperability using automation principles. Orchestrated: Take manual processes out of the equation and introduce end to end orchestration of the network from device-to-the-cloud coordinated for a given set of network connectivity, services, and performance objectives. In such a world you will have enable cloud applications to truly work better across programmable networks and we will be ready to deal with scale, complexity, security as the Digital Economy evolves and matures. Smart Cities Device Residential Network Agile, Assured, Orchestrated
8
Applications driving programmable networks Application Intent APIs
Orchestration L4-L7 Services Security Services IP Services Carrier Ethernet Services Optical/Wavelength Services So what do we need to work towards that vision. At the very heart of it – a means for applications and networks to be more aware of each other, for applications to be able to directly communicate their requirements to the network, the network automatically orchestrating its services & constantly providing fine-grained actionable visibility end-to-end from device to cloud back to the application. Services
9
Applications driving programmable networks
Multi-Media Services Cloud SaaS Applications Video Surveillance Video UC&C VTC AR/VR VoIP WebRTC Cloud Platform Micro Services Platform Cloud Core DCs Cloud Regional DCs Cloud POP DCs Cloud Central Office DCs Cloud Edge DCs Elastic Bandwidth Low Delays, Packet loss, Jitter Application Intent APIs () Programmable Network Control Orchestrator Lets think about the high-level functional components in such an architecture At the very bottom we have the physical and virtual networks elements that provide the end-to-end connectivity between the device end-points and the cloud applications. The layer above that is the Network/Service level orchestrator that can program the underlying data plane for unique requirements of the Cloud-hosted applications that sit on top on a Cloud Platform. The Cloud platform itself could span multiple domains going from core, to regional, POP, CO and to the edge with the application running as a collection of loosely coupled services across one or more of these domains. Regardless of where the application micro service is running it should be able to communicate requirements such as – low delay, latency, jitter, elastic bw. which the orchestrator should be able to provision by programming the underlying DP using SDN/NFV or Legacy interfaces. You have agility, assurance, orchestration right there. SDN Legacy Networks NFV SBIs SBIs SBIs Programmable Network Elements PNFs Legacy VNFs
10
RTM Session Event Network Elements Events:
Lync 2014 9/8/2018 RTM Session Event Events: Start, Update, Error & End Dialog Attributes: SIP URIs Call ID IP SA/DA Protocol Transport S/D Ports Media Type Codec & BW RTM RTM Session Event Application Intent APIs () Programmable Network South Bound API To illustrate our case lets look at Real-time Media where its on-prem deployments today experience intermittent session quality and reliability issues. Now amplify the complexity as the real-time applications migrate to the cloud using some intermediary set of networks, managed by multiple different providers to get to the real-time application services. On this slide you see Microsoft Skype for Business SDN Interface that provides a subscription-based interface for network controllers or network management systems to receive call data to monitor and analyze network traffic in order to optimize the Skype for Business media stream quality. Switch IDS/IPS Firewall Probe WAN Optimizer Router WLAN AP RTM RTM Network Elements Signaling Media Source: Microsoft Skype for Business © 2014 Microsoft Corporation. All rights reserved. Microsoft, Windows, and other product names are or may be registered trademarks and/or trademarks in the U.S. and/or other countries. The information herein is for informational purposes only and represents the current view of Microsoft Corporation as of the date of this presentation. Because Microsoft must respond to changing market conditions, it should not be interpreted to be a commitment on the part of Microsoft, and Microsoft cannot guarantee the accuracy of any information provided after the date of this presentation. MICROSOFT MAKES NO WARRANTIES, EXPRESS, IMPLIED OR STATUTORY, AS TO THE INFORMATION IN THIS PRESENTATION.
11
RTM Quality Event Network Elements Events:
Lync 2014 9/8/2018 Events: Quality Update event for Voice, Video or Data Attributes: 5 Tuple Value NMOS Value RTM Delay Value RTM Jitter Value RTM Packet Loss Value Healer Ratio Value RTM Quality Event UC UC Quality Event Application Intent APIs () Programmable Network South Bound API Here you see the MS SfB SDN Interface providing Quality Data to SDN Controllers and NMS. Metrics like delay, jitter, packet loss can be used as indicators by the SDN Controller to trigger the identification of failures or bottlenecks in the network that it can then correct by reprogramming alternate paths for the media stream. Switch IDS/IPS Firewall Probe WAN Optimizer Router WLAN AP UC UC Network Elements Signaling Media Source: Microsoft Skype for Business © 2014 Microsoft Corporation. All rights reserved. Microsoft, Windows, and other product names are or may be registered trademarks and/or trademarks in the U.S. and/or other countries. The information herein is for informational purposes only and represents the current view of Microsoft Corporation as of the date of this presentation. Because Microsoft must respond to changing market conditions, it should not be interpreted to be a commitment on the part of Microsoft, and Microsoft cannot guarantee the accuracy of any information provided after the date of this presentation. MICROSOFT MAKES NO WARRANTIES, EXPRESS, IMPLIED OR STATUTORY, AS TO THE INFORMATION IN THIS PRESENTATION.
12
Intent Logical Concepts
13
Automation with Life-cycle Services Orchestration
What you saw earlier was an on-prem scenario. Now what happens when you migrate the applications to the Cloud. Multiple SP domains between the device and the cloud. How do you orchestrate and program application requirements across these multiple domains? That’s where MEF’s work on Lifecycle Services Orchestration comes into the picture. A single entity called the LSO sits atop SP domain infrastructure. Today on the north, it integrates with SP applications like OSS/BSS and on the South it works with SDN/NFV/WAN/Traditional NMS to program the underlying infrastructure. It also talks to its peers managing other SP domains to ensure end-to-end connectivity services that come with capabilities like Performance, Assurance, Fulfillment, Analytics and Policy. Tomorrow we will be expanding the LSO coverage to not-only integrate with SP applications but also Enterprise Cloud Applications such as RTM. This has the promise of providing end-to-end QoE, visibility and diagnostics as UC moves to the Cloud also known as UCaaS.
14
Network automation API stack
Application Intent NBIs Orchestration SaaS Applications Software Programmable Networks Physical and Logical Network Elements SBIs
15
Current LSO Model Lifecycle Service Orchestration Controller
Fulfillment Performance Policy Control Assurance Usage Security Analytics Controller NFV Orchestration Here is the current LSO Model that can orchestrate multiple SP domains with L2 and L3 connectivity services that come with capabilties such as fulfillement, performance, analyics. SBIs SBIs PNFs VNFs
16
Proposed LSO Model with Application Intent APIs
Real-Time Media Apps Application Intent APIs Lifecycle Service Orchestration Fulfillment Performance Policy Control Assurance Usage Security Analytics SDN Controller NFV Orchestration In the future we will expand that to include open and standards-based Application Intent APIs to provide services across the networking stack based on the requirements of SaaS applications like RTM. SBIs SBIs PNFs VNFs
17
Application Programming Model Examples
Programable Network App On Device Cloud Service Telemetry ST Data LSO Intent NBI Inference Intent Model ST App & Proxy Intent API Analytics Direct Application Intent Model Application Proxy App Intent API Analytics Programable Network App On Device Cloud Service Telemetry Application Data LSO Intent NBI DPI Intent Model DPI App & Proxy DPI Intent API Analytics Telemetry Application Data Telemetry Application Data LSO Intent NBI App On Device Cloud Service Programable Network Not all applications will expose APIs like what we say with Microsoft SfB SDN APIs. In this slide you see 3 different application programming models Direct model where application programs the networks directly via LSO using Intent APIs. Existing applications who have this capability and new applications that will expose Intent APIs will be a good model for this. DPI Intent Model where a DPI engine acts as a proxy for the application. It uses DPI-data to gather insights about the application and programs the network underneath via LSO using Intent APIs. Legacy applications that do not have Intent APIs can leverage this model Interface Intent Model where you inject synthetic traffic alongside the application stream to make a second-degree assessment of the application quality based on the performance data collected for the synthetic steam. E.g.
18
LSO Closed Loop Analytics with Application APIs
Framework Network Telemetry Application Telemetry Cloud Applications Network Events Application SOF LEGATO (BUS:SOF) Service Orchestration Functionality AI Analytic Applications Service Orchestration Functionality Data Analytics Platform PRESTO (SOF:ICM) Infrastructure Control and Management In the previous slides you see how Analytics is one of the capabilities offered via LSO. Analytics is what provides the end-to-end visibility and diagnostics necessary to troubleshoot under SaaS applications underperform. This slide shows how. LSO has a data analytics platform that is able to collect network as well as application telemetry data into a big data lake. LSO also has analytics application that mines the data to generate real-time insights that can be programmed back into the network via LSO. The trigger for the Analytics Application is usually event-driven with events coming both from the Network (network failure) and from the application (quality event) ADAGIO (ICM:ECM) Element Control Management & Data Aggregation Events Logs Metrics Netflow Network Infrastructure
19
Use Case: Automating QoE using RTM-SDN
Orchestrator AI Analytic Capabilities SDN Controller At the IMTC they have been driving 2 specific use cases for RTM – Automating QoE, Automating Diagnostics. On this slide you see the capabilities that an LSO would need to provide – Dynamic QoS marking, admission control, dynamic TE, policy for QoE automation. These are well-defined in a white paper with the link on AQS on the IMTC website. Automated QoE Service (AQS) Automated QoS Admission Control Dynamic Traffic Engineering (SD-WAN)
20
Use Case: Automating Diagnostics using RTM-SDN
Orchestrator AI Analytic Capabilities SDN Controller On this slide you see the capabilities that an LSO would need to provide for automating diagnostics – Session monitoring, Data Collector (from app and from network), analytics on the data, as well as actionable visibility. These are well-defined in a white paper with the link on ADS on the IMTC website. Automated Diagnostics Service (ADS)
21
Example: AQS & ADS for Cloud-hosted RTM
This is the IMTC usecase for UCaaS with RTM in the Cloud. With ADS and AQS as services in the LSO (or SDN Controller) the RTM application in the Cloud will be able to communicate Intent for the desired QoE and Diagnostics end-to-end. SDN RTM Cloud Service (Cloud Hosted RTM) Extensions for Multi-domain, Multi-tenancy, Carrier-scale
22
Cisco Reactive Networking
Source: Cisco ONS 2016 Keynote on Reactive Networking
23
Google Project Espresso
Here is another real-world deployment of Application Intent developed by Google Project Expresso. Google globally has about 200 peering points across the globe. At each peering point, they have replaced expensive peering routers with commodity switching h/w that implements its own fully-compliant BGP speaker. A local Controller at each node programs every application packet with a label. It also programs the switching fabric to egress a packet a certain port based on its label. So all packet forwarding intelligence resides on the local controller and the switching fabric simply forwards packets as programmed. The Local Controller receives its forwarding intelligence from a Global Controller that is collecting application performance signals from all the peering sites across the globe. It has statistics that says how well an application has performed when using 1 egress port vs another in other words when peering with 1 network versus another. The Global controller uses this information to reprogram the forwarding rules into the local controllers as necessary.
24
Microsoft On-Premise Skype for Business
9/8/2018 Microsoft On-Premise Skype for Business Drive Agility and Automation into Networks Diagnostics Proactively automate the diagnostics and troubleshooting of poor quality real-time media calls Automated QoS Reduce the cost of deploying QoS and implementing Traffic Engineering. Minimize operational complexity Orchestration IT Agility & Network Customization through real-time, policy-based config of network resources and devices. It supports the real-time policy-based configuration of network resources and devices when deploying SfB on-prem It supports the automated QoE usecase – with TE It supports the proactive diagnostics for ease of troubleshooting with poor call quality. Source: Microsoft Skype for Business © 2014 Microsoft Corporation. All rights reserved. Microsoft, Windows, and other product names are or may be registered trademarks and/or trademarks in the U.S. and/or other countries. The information herein is for informational purposes only and represents the current view of Microsoft Corporation as of the date of this presentation. Because Microsoft must respond to changing market conditions, it should not be interpreted to be a commitment on the part of Microsoft, and Microsoft cannot guarantee the accuracy of any information provided after the date of this presentation. MICROSOFT MAKES NO WARRANTIES, EXPRESS, IMPLIED OR STATUTORY, AS TO THE INFORMATION IN THIS PRESENTATION.
25
Long Term Vision of Ecosystem Automation
SaaS Application APIs SaaS Applications Self-service Web Portal East-West APIs East-West APIs East-West APIs Presto APIs Presto APIs Presto APIs NFV MANO Equipment Equipment NFV MANO SDN Controller SDN Controller NFV MANO Cloud Orchestrator (CLO) SDN Controller VM UNI ENNI ENNI UNI SDN Switch Access Provider Business Cloud Provider Enterprise Network Retail Provider Third Network Services East-West APIs North-South APIs
26
The Evolution of Managed SD-WANs
Application Performance Assurance Application Services L4-L7-as-a-Services Security-as-a-Service IP VPN Tunnel Mobile Broadband – CE – TDM – Broadband L4-7 Point Service Management Security Management APIs Overlay Performance Management Physical Underlay Management
27
The MEF 3.0 Overlay Virtualized Model
Cloud APIs Self-service Web Portal Cloud Orchestrator (CLO) East-West LSO Sonata and LSO Interlude APIs VM Mission Critical Cloud Applications LSO Presto APIs LSO Presto APIs UNI NFV MANO SDN Controller SD-WAN Controller CPE Managed Networks (CE, MPLS,5G) SDN Switch Load Bal. SD-WAN Internet ISP BGP SD-WAN Proposed SD WAN Interaction with LSO SOF via LSO Presto Fulfillment (v)CPE provisioning (location, UNI configuration) Path Selection (Application mapping to link or bundle, app priority, link performance reqs, encrypted vs. clear) Security (provision operators, Radius/Tacacs/SAML for enterprise authentication) Routing (both LAN side as well as NNI and forwarding node selection) Segmentation (for security or for downstream classification) Analytics Flow statistics (users, application, destinations) Usage per tenant & CPE Assurance Health checks of SDWAN nodes (CPE, Gateways, RO) and overlays (VPN, bundle) Performance Link status & statistics (underlay) Application quality improvements (MOS, Transaction time) UNI UNI APM IPS/IDS Firewall Carrier NAT BGP Routing Service Provider SDN Switch GW RO - Resource Orchestrator GW - Gateway CPE - Customer Premise Equipment APM – Application Performance Management IPS/IDS – Intrusion Prevention System / Intrusion Detection System
28
Application Committee
29
Applications Committee Mission
As more and more enterprises move their applications to the cloud, there is a need to ensure that cloud applications deliver a high quality, frictionless end-user experience, as if their applications were hosted locally in their own organizational sites. The objective of the Applications Committee is to ultimately enable applications to dynamically interact with the network using a set of machine-to-machine APIs, thereby ensuring that application-level quality of experience, security, and performance requirements can be met by the underlying network infrastructure without human intervention. This new work completely aligns with the three pillars of MEF's Third Network vision: Agile – Enable networks to be programmed by applications for a given set of autonomic behaviors such that control, visibility, and performance guarantees can be realized in an on-demand and almost real-time manner. Assured – Enable networks to deliver on application performance objectives, availability, security, and middle box interoperability using automation principles. Orchestrated – Enable networks that are inter-connected from device-to-the-cloud to be orchestrated, automated, and coordinated for a given set of network connectivity, services, and performance objectives.
30
Application Committee Work
The Application Committee will kick off with the active work of the IMTC, including that of the Real-Time Media (RTM SDN) working group and other IMTC work areas (see imtc.org for additional information about IMTC). Application sub-groups will deliver various automation use cases, building upon those already published by IMTC. MEF and IMTC have entered into an asset transfer agreement whereby all assets of IMTC will be transferred to MEF and active work will be carried on within MEF. IMTC member companies that are not already members of MEF will become MEF principal members - approximately 10 in number. The agreement has been approved by both Boards of Directors and the IMTC membership, and requires final regulatory approval, which is expected to be complete by end of October of this year. To support a higher level of abstraction interaction between applications and networks, an Intent Work Area will initially start under the Application Committee to collaboratively develop a common, multi-vendor, interoperable, intent-based set of APIs enabling networks to be treated as a programmable abstraction layer comprised of a closed-loop, black box model.
31
Application Committee Org Chart
Real Time Media (IMTC AGs) RTM SDN AG Co-Chairs UC IoT Cloud Services IoT Intent Group Area Directors Chairs (Interim Pascal Menezes) David Lenrow John Strassner Chris Lauwers Pascal Menezes Darren Gallager Mani Mahalingam Rao Chandrashekar
32
Preliminary Commitments
Call for Chairs Preliminary Commitments Drive Application Committee groups to deliverable goals and milestones coordinated with the CTO and MEF Technical Leadership team Attend most MEF F2F meetings and facilitate agenda and group work items Commit to about 4-8 hours a week Send inquiries to Kevin Vachon
33
Lifecycle Service Orchestration
The New Work of the MEF Applications Lifecycle Service Orchestration Specifications Reference Implementations Certifications Marketing Programs Services
34
Thank You
Similar presentations
© 2025 SlidePlayer.com. Inc.
All rights reserved.