Publications

Network Monitoring for SDN Virtual Networks

IEEE INFOCOM 2020 - IEEE Conference on Computer Communications, Acceptance rate 19.8%, Virtual (originally in, Toronto, ON, Canada), 2020, pp. 1261-1270, doi: 10.1109/INFOCOM41043.2020.9155260.

Gyeongsik Yang, Heesang Jin, Minkoo Kang, Gi Jun Moon, Chuck Yoo

Download PDF

Abstract

This paper proposes V-Sight, a network monitoring framework for software-defined networking (SDN)-based virtual networks. Network virtualization with SDN (SDN-NV) makes it possible to realize programmable virtual networks; so, the technology can be beneficial to cloud services for tenants. However, to the best of our knowledge, although network monitoring is a vital prerequisite for managing and optimizing virtual networks, it has not been investigated in the context of SDN-NV. Thus, virtual networks suffer from non-isolated statistics between virtual networks, high monitoring delays, and excessive control channel consumption for gathering statistics, which critically hinders the benefits of SDN-NV. To solve these problems, V-Sight presents three key mechanisms: 1) statistics virtualization for isolated statistics, 2) transmission disaggregation for reduced transmission delay, and 3) pCollector aggregation for efficient control channel consumption. V-Sight is implemented on top of OpenVirteX, and the evaluation results demonstrate that V-Sight successfully reduces monitoring delay and control channel consumption up to 454 times.

Paper summary

Monitor virtual networks without duplicating collection

Virtual-network measurements can share physical collection work while remaining separate for each tenant.

V-Sight addresses three monitoring problems in SDN virtual networks: counters that mix tenants, delays while gathering statistics, and repeated control-channel traffic.

Translate counters and share collection work

Statistics virtualization derives tenant-specific counters from physical measurements and the virtual-to-physical resource map. Transmission disaggregation collects physical statistics ahead of tenant requests, so a controller does not have to wait for every physical switch when it asks for a counter. Compatible collectors are then merged to reduce repeated control messages. Their timing is adjusted to keep the cached values useful for the different request intervals.

How V-Sight combines tenant isolation with shared monitoring work.
MechanismRole
Statistics virtualizationDerive tenant counters from physical measurements
Transmission disaggregationCollect before tenant requests arrive
Collector aggregationMerge compatible collection requests

Balance faster replies against data freshness

In linear and fat-tree test networks, V-Sight reduces statistics transmission delay and control-channel traffic relative to the OpenVirteX-based virtualized baseline. The largest measured reply delay with V-Sight is below 20 ms, although it remains higher than native SDN in the comparisons. Cached measurements have a time difference between collection and the tenant’s request. The reported average differences stay within 2.5 seconds for the tested five-second request interval.

Control-channel traffic for NV, V-Sight and native SDN in linear and fat-tree networks.View full-size figure
V-Sight uses less monitoring traffic than the virtualized NV baseline in both topologies. Native SDN is shown as the non-virtualized reference.

Key takeaways

  1. Tenant isolation can coexist with shared measurement collection. V-Sight shares physical collection while translating its results into separate tenant statistics.
  2. Faster replies and fresher measurements are different goals. Cached counters reduce repeated requests, with freshness determined by when those counters were collected.