# Flexibility Matters

The emerging trend to softwarize networks based on concepts such as Network Virtualization, Software Defined Networking (SDN) and Network Functions Virtualization promises to increase flexibility in networking. So far, flexibility is used rather as a qualitative argument for a network design choice. Furthermore, the meaning of flexibility behind such qualitative arguments is highly varying in the state of the art as a common understanding of flexibility is missing.

We present an approach towards evaluating network flexibility through a definition of a flexibility measure, which provides a quantitative analysis and a comparison of different network designs. Further, it can illustrate the trade-off between flexibility and cost needed to provide flexibility.

This web page intends to collect discussions and results about network flexibility including definitions, guidelines, benchmarks, use cases and tools. Stay tuned!

# Flexibility Measure (Proposal)

Network flexibility refers to the ability to support adaptation requests (=challenges) that can be, for example, changes in the requirements or new traffic distributions. We define the flexibility $\rho$ of a system $S$ as the fraction of requests that can be supported from a given set and sequence of requests and within a given time threshold $T$ and cost budget $C$ for each adaptation request:

$$\rho_{T} = \frac{\mid \text{supported requests within } T \text{ and } C\mid}{\mid \text{total number of requests}\mid}$$

# Flexibility Aspects

As with other network measures like Quality of Service, Flexibility as a measure is not a singular measure, but used with a specific objective in mind. To reflect such objectives, we come up with flexibility aspects. A flexibility aspect describes a concrete ability in which a network can adapt, e.g., change the flow routes or change of the allocated resources.

Category Aspect
Function Configuration
Parameter Configuration
Locate Functions Function Placement
Scale Resource and Function Scaling

# Use Case Examples

### Dynamic SDN Controller Placement

A dynamic SDN control plane can adapt its configuration to changing traffic flows to achieve an optimal con- trol performance, i.e., minimal average flow setup time. The flexibility of the dynamic SDN control plane represents its ability to respond to changing traffic flows under a migration time threshold $T$.

Intuitively, one would think a 4 controller system is more flexible than a 1 controller system. However, as we show in [3] for a very tight adaptation time threshold $T$, a 1 controller system exhibits higher flexibility.

Objective Aspect Request Measure Time threshold Cost
Control Performance of dynamic SDN control plane Function placement Flow arrival (from distribution) Fraction of successful controller placements Varied Average flow setup time
Failure recovery of SDN system Flow configuration All possible single and dual link failures Fraction of recovered failures Varied resource overhead for reservation

### Failure Recovery in Software Defined Networks

The flexibility of survivable networks can be defined as the ability to recover from failures in a timely manner. Network operators can choose either proactive recovery approaches (i.e., protection) where the network is prepared for failures in advance at the cost of redundant resources; or reactive approaches (i.e., restoration).

As we show in [1], networks with restoration are less flexible for very tight time thresholds $T$ but can supersede networks with protection for larger $T$.

# Selected Publications

• [1] W. Kellerer, A. Basta, P. Babarczi, A. Blenk, M. He, M. Klugel, A. Martinez Alba: How to Measure Network Flexibility? A Proposal for Evaluating Softwarized Networks. IEEE Communications Magazine, 2018 PDF
• [2] A. Blenk, P. Kalmbach, J. Zerwas, M. Jarschel, S. Schmid, W. Kellerer: NeuroViNE: A Neural Preprocessor for Your Virtual Network Embedding Algorithm IEEE INFOCOM 2018 (main conference), Honolulu, HI, USA, April 15‐19, 2018. PDF
• [3] M. He, A. Basta, A. Blenk, W. Kellerer: How Flexible is Dynamic SDN Control Plane? Software-Driven Flexible and Agile Networking (SWFAN), IEEE INFOCOM Workshop, 2017. PDF
• [4] A. Basta, A. Blenk, K. Hoffmann, H.-J. Morper, M. Hoffmann, W. Kellerer: Towards a Cost Optimal Design for a 5G Mobile Core Network based on SDN and NFV. IEEE Transactions on Network and Service Management 14 (4), 2017, 1061 – 1075. PDF
• [5] W. Kellerer, A. Basta, A. Blenk, Using a Flexibility Measure for Network Design Space Analysis of SDN and NFV, SWFAN’16, IEEE INFOCOM Workshop, 2016. PDF
• [6] A. Blenk, A. Basta, J. Zerwas, M. Reisslein, W. Kellerer: Control Plane Latency With SDN Network Hypervisors: The Cost of Virtualization. IEEE Transactions on Network and Service Management 13 (3), 2016, 366- 380. IEEE Xplore
• [7] W. Kellerer, A. Basta, A. Blenk, Flexibility of Networks: a new measure for network design space analysis?. arXiv preprint arXiv:1512.03770, 2015. PDF