2023 Fiscal Year Research-status Report
Data Plane Programming Approach for Real-time Failure Detection and Localization
| Project/Area Number |
23K16875
|
|---|
| Research Institution | Kyoto College of Graduate Studies for Informatics |
|---|
Principal Investigator |
望月 バドル 京都情報大学院大学, その他の研究科, 准教授 (10838460)
|
|---|
| Project Period (FY) |
2023-04-01 – 2026-03-31
|
| Keywords | Inband Network Telemetry / P4 / SDN / Network Failure |
|---|
| Outline of Annual Research Achievements |
The proposed research is dedicated to network failure detection and localization in optical core networks. The aim is to leverage the low latency of data plane programmable network devices and In-band Network Telemetry (INT) for both detection and precise localization in real-time of network failures. This research is to be performed in three years under three phases:
1: Precise localization of network failures
2: Real-time detection of network failures
3: Dynamic orchestration of INT metadata collection
In this first year, Phase 1 has been achieved and research work on phase 2 has started. In the first year, a precise failure localization algorithm has been developed. It uses backward probing in order to determine the exact switch and link where the network failure has occurred.
|
| Current Status of Research Progress |
Current Status of Research Progress
2: Research has progressed on the whole more than it was originally planned.
Reason
The research is having a smooth progress: one journal paper on precise failure localization algorithm is under submission and the development of phase 2 has started.
|
| Strategy for Future Research Activity |
We will continue research work on phase 2 and phase 3 as follows.
Phase 2: Real-time detection of network failures. Our main objectives will be to: 1) Set up path tables and 2) Detect failure through timed-out path table entries
Phase 3: Dynamic orchestration of the collection of INT metadata. Our main objectives will be to: 1) Maximize the number of collected INT metadata and 2) Identify which INT metadata is important to be collected
After phase 3 is finished, we will integrate the modules of all three phases into a global framework and investigate its performance.
|
| Causes of Carryover |
The article “MAC computer”, purchased in order to perform numerical simulations for the proposed research required less than the expected costs.
The remaining amount will be used to buy additional Rasberry Pis and a dell tower computer in order to perform larger experiments to validate simulation results.
|
