SLIDE 1
Env2Vec
Accelerating VNF Testing with Deep Learning
Eurosys 2020 Guangyuan Piao, Pat Nicholson, Diego Lugones
Customer Network Code Build Deploy Release Operate VNF Developer
Env2Vec Accelerating VNF Testing with Deep Learning Guangyuan Piao, - - PowerPoint PPT Presentation
Env2Vec Accelerating VNF Testing with Deep Learning Guangyuan Piao, - - PowerPoint PPT Presentation
Env2Vec Accelerating VNF Testing with Deep Learning Guangyuan Piao, Pat Nicholson, Diego Lugones Eurosys 2020 Release Operate Deploy Build Code VNF Developer Customer Network Release Operate Deploy Build Code New features VNF
SLIDE 2
SLIDE 3
Customer Network Code Build Deploy Release Operate VNF Developer
New features
SLIDE 4
Customer Network Code Build Deploy Release Operate VNF Developer VNF testing ü 5 9’s ü SLO’s ü Assurance ü …
New features Upgrades
time-consuming
SLIDE 5
VNF testing KPI’s
?
Middleware OS Virt/Container HW VNF
Config
Cloud stack
SLIDE 6
VNF testing
?
KPI’s Anomaly !
ML
Middleware OS Virt/Container HW VNF
Config
Cloud stack
SLIDE 7
high-dimensional parameter space
Middleware OS Virt/Container HW VNF
Config
Cloud stack
Middleware OS Virt/Container HW VNF
Config
Different stack
SLIDE 8
Env2Vec
Accelerating VNF Testing with Deep Learning
1) Robust to environment variations 2) Simple single ML model 3) Work in previously unseen environments
Middleware OS Virt/Container HW VNF
Config
Middleware OS Virt/Container HW VNF
Config
SLIDE 9
Deep Learning Architecture
Env2Vec
Accelerating VNF Testing with Deep Learning
Anomaly detection
Embeddings
Environment Metadata
Middleware OS Virt/Container HW VNF
Config…
T r a i n i n g
FNN
Contextual Features
FNN: FeedForward Neural Network
GRU
GRU: Gated Recurrent Units
Historical data
SLIDE 10
@github: WillKoehrsen/wikipedia-data-science
Book (wikipedia) embeddings by genre Environments Embeddings by test case
Build type D (debug), T (test), S (stable), etc.
SLIDE 11
Env2Vec
Accelerating VNF Testing with Deep Learning
Anomaly ! Middleware OS Virt/Container HW VNF
Config
KPI’s timeseries
R u n t i m e
Deep Learning Architecture
Anomaly detection
Embeddings
Environment Metadata
Middleware OS Virt/Container HW VNF
Config…
T r a i n i n g
FNN
Contextual Features
GRU
Historical data
SLIDE 12
KDN dataset (public)
Open Virtual Switch Snort SDN-enabled firewall
Carrier-grade VNF for multiple
Testing environments Build types Services Under test
Accuracy 86.2% - 100% False alarms reduced by 20.9% to 38.1% Simplified adoption: single model competitive against multi-model proposals
SLIDE 13
Env2Vec
Accelerating VNF Testing with Deep Learning
Eurosys 2020 Guangyuan Piao, Pat Nicholson, Diego Lugones