The emergence of Software Defined Wide Area Networks (SD-WAN) has provided a cost-effective and flexible alternative to the expensive and rigid Multiprotocol Label Switching (MPLS) technology of earlier years. Wide Area Network (WAN) optimization techniques and appliances were developed in the early 2000s to address the issues in a WAN implemented with MPLS. The widespread adoption of SD-WAN raises the question of whether WAN optimization is being phased out as MPLS declines. Let us examine whether that is indeed the case.
The main benefit of WAN optimization is a reliable Quality of Service (QoS) even under peak traffic conditions on the network. WAN optimization appliances achieved this through preserving bandwidth, reducing latency and alleviating packet loss.
- Bandwidth Preservation: As the data traffic increased, WAN optimization managed to preserve throughput using data deduplication and compression. It prioritized audio/video data so that network congestion did not negatively affect time-sensitive applications like Voice over IP.
- Network Latency Reduction: Another impact of increasing data traffic is an increase in network latency, i.e., the increase in time to deliver a data packet from source to destination. WAN optimization appliances addressed network latency issues through data caching and transport protocol specific optimizations.
- Packet Loss Alleviation: A packet loss rate of about 1 percent is common for internet connections due to traffic congestion or network failures. WAN optimization used Forward Error Correction (FEC) techniques to recreate missing packets, thus avoiding the delays inherent in resending the lost data packets.
But the growth of the internet and the emergence of the cloud made WAN optimization increasingly complex and costly, which paved the way for the development and widespread adoption of SD-WAN.
Benefits of SD-WAN
Enterprise businesses found SD-WAN attractive because of its low cost, ease of use, and flexibility. It also attempted to provide the performance guarantees of WAN optimization.
- Lower Cost: Use of internet circuits makes SD-WAN less expensive than MPLS circuits. Moreover, SD-WAN creates a virtual overlay over disparate transport protocols like Cable, DSL, Fiber and 4G/LTE. This technique further reduces costs and makes SD-WAN more flexible than MPLS.
- Increased Capacity and Performance: SD-WAN offers much greater capacity since it uses the internet. The availability of a large number of inexpensive internet circuits facilitates the aggregation of several internet connections for load balancing, thus improving redundancy and performance. SD-WAN also uses Policy-based Routing (PbR) to improve performance; PbR is used with monitoring of both network latency and packet loss to select the optimal transport for application-specific traffic.
- Cloud Suitability: MPLS has a disadvantage in accessing the cloud since it is optimized only for traffic from branch offices to the data center, but not for the branch to cloud traffic. Since SD-WAN was designed for internet circuits, it is inherently suitable for the cloud.
Although SD-WAN offers many beneficial features, it is not able to match the QoS guarantee that WAN optimization can provide. This is because SD-WAN does not have the capability to control the data packets after they have left the source; it simply selects the best performing WAN to use for the application data but has no mechanism to improve the WAN performance as traffic conditions change.
This gap can only be resolved by making SD-WAN work with WAN Optimization. Cato Networks Secure, Global SD-WAN as a Service is a ‘best of both worlds’ solution that uses the benefits of both SD-WAN and WAN optimization. It provides network security as well. The Cato solution provides a cloud-based service with secure SD-WAN, QoS thorugh SLA-backed connectivity, and extensions of the WAN to cloud-based data centers and mobile users. With this all in one solution, customers can realize lower costs by eliminating multiple endpoint legacy products that provided the services separately.