Project Technical Lead: Adnan Saleem. Elected 1/17/19.
Project Committers detail:
Initial Committers for a project will be specified at project creation. Committers have the right to commit code to the source code management system for that project.
A Contributor may be promoted to a Committer by the project’s Committers after demonstrating a history of contributions to that project.
Candidates for the project’s Project Technical Leader will be derived from the Committers of the Project. Candidates must self nominate by marking "Y" in the Self Nominate column below by Jan. 16th. Voting will take place January 17th.
Only Committers for a project are eligible to vote for a project’s Project Technical Lead.
...
Latest documentation:
Source code available on Gerrit: https://gerrit.akraino.org/r/admin/repos/sdt
Mailing list: sdt-blueprint@lists.akraino.org
Project Technical Lead: Colin Peters (Colin Peters elected )
Project Committers detail:
Committer | Committer Company | Committer Contact Info | Committer Bio | Committer Picture | Self Nominate for PTL (Y/N) |
Prakash Siva
Radysis
| Fujitsu | iino.hatsumi@fujitsu.com | ||||
| Fujitsu | t_fukuda@fujitsu.com | ||||
| Fujitsu | shigetome.hiroj@fujitsu.com | ||||
| Fujitsu | inoue.reo@fujitsu.com | ||||
| Fujitsu | fukano.haruhisa@fujitsu.com | ||||
| Yasushi Kurokawa | Fujitsu | kurokawa.yasu@fujitsu.com | |||
| Fujitsu | tsuji.yoshiko@fujitsu.com | ||||
| Colin Peters | Fujitsu | colin.peters@fujitsu.com | Y | ||
| JFTT | zhaomin@fujitsu.com | ||||
| Guixiang Miao | JFTT | miaoguixiang@fujitsu.com |
Presentation:
| View file | |||||||||
|---|---|---|---|---|---|---|---|---|---|
|
Use Case Details:
Attributes | Description | Informational |
Type | New |
Industry Sector |
Telco Carrier Networks and Enterprises
| CPS, IoT | |
Business driver |
Vast amounts of mobile/wireline data (predominantly video) is expected to continue to grow, particularly with 5G and IoT. Low latency, backhaul bandwidth restrictions/cost, and real time edge media analytics require media processing at network edges versus transporting all media to network core. Without the ability to process real time media at the network edges a number of new advanced applications would not be possible nor economically viable.
Business use cases
- Edge deployments at enterprises, entertainment venues, factory automation plants, public facilities where real time media processing required
- Edge media applications include multi-party conferencing, gaming, surveillance, IoT generated content, AR and VR applications
- Edge media applications requiring low latency and to overcome backhaul BW availability and costs being prohibitive
- Real time media analytics with AI and ML based applications for high value and media monetization applications
| Cyber physical systems which combine sensor network with computing to monitor and control the physical environment become popular. The bandwidth of the sensor network depends on use cases. Large amounts of data from large amounts of sensor nodes will pressure the NW bandwidth between edge and clouds. Therefore, we need to have a means to optimize each NW bandwidth according to use cases. This blueprint propose a solution for NW bandwidth optimization. | ||
Business use cases | Smart city, agriculture, interactive live sports | |
Business Cost - Initial Build Cost Target Objective |
Initial build requires a small footprint POD with minimal fabric and management switch, 4+ compute nodes with optional GPU acceleration, local storage node(s), PSUs, rack, typically under $100K with SW
Security need
POD platform SW and application level security vulnerability scanning and automated patching capabilities required
Media content security and user access authentication capabilities required
Regulations
Depending on type of Edge Media application GDPR or other regulatory requirements may be applicable. NEBS may be required depending on deployment location and carrier network requirements
| Depends on use cases. E.g. Monitoring sewerage water level Gateway:$3000 Sensor node:$2500 Water level sensor:$1500 | ||
Business Cost – Target Operational Objective |
- Low operation cost, with support for remote FCAPS management, and ONAP based zero-touch resource and service orchestration
- Typical 16U height OCP rack with similar power consumption, with minimal footprint of 2 compute nodes.
- Edge Media solution shall support POD level consolidated management (OSAM) and service level orchestration and LCM via ONAP.
- Zero touch provisioning, upgrades, fault and performance management KPI, and auto-scaling and auto-healing capabilities
| Depends on use cases. ・Power consumption and management for sensor node and gateway ・Cloud etc… | ||
Security need | The sensor node and gateway will be used outdoors in untrusted environment and it handles potentially privacy-sensitive data such as live video. Therefore, the device needs to support trusted boot, trusted key storage, and encrypted communication. | |
Regulations | Depends on use cases. E.g. Monitoring sewerage water level There are several environmental design guidelines.(IPx7, etc..) | |
Other restrictions | Depending on |
Additional details
Edge Media solution shall enable support for high density media processing via GPU or FPGA acceleration for advanced high density AI and ML applications and shall scale from single site to 100s in regional deployments to 1000s globally
Additional details on architecture and use cases documented in supplementary PPT
Case Attributes
Description
Informational
Type
New
Blueprint Family - Proposed Name
Network Cloud, RT Cloud
Use Case
Real Time Edge Media Processing
Blueprint proposed
- Unicycle POD (4-6 servers, single 16U rack configurations)
- Tricycle POD (16U or 42U rack configurations, multi-rack)
- Cruiser POD (Multi-rack Core Network Configurations, with spine leaf fabric and ToR switch)
Initial POD Cost (capex)
Estimates (TBD)
- Unicycle POD (< 100K)
- Tricycle POD (< 200K)
- Cruiser POD (< 300K)
Scale
- Unicycle POD – 1 rack with < 6 servers
- Tricycle POD – Multiple racks, each with < 24 servers
- Cruiser POD – Multiple racks, each with < 96 servers
Applications
Edge Virtual Function Applications (reference)
- Edge deployments at enterprises, entertainment venues, factory automation plants, public facilities where real time media processing required
- Edge media applications include multi-party conferencing, gaming, surveillance, IoT generated content, AR and VR applications
- Edge media applications requiring low latency and to overcome backhaul BW availability and costs being prohibitive
- Real time media analytics with AI and ML based applications for high value and media monetization applications
Power Restrictions
TBD
Preferred Infrastructure Orchestration
OS – CentOS or similar Linux, KVM
Under Cloud – Airship
OpenStack – VM Orchestration
Docker + K8S - Container Orchestration
VNF Orchestration - ONAP
SDN
OVS-DPDK, SR-IOV
Workload Type
VMs, Containers
Additional Details
Edge Media solution shall enable support for high density media processing via GPU or FPGA acceleration for advanced high density AI and ML applications.
| use cases, there can be other requirements. | ||
Additional details | NA |