By Erik Fritzler, Contributor, Network World
As an ever present staple of technology in the inventory of any major enterprise, WAN optimizers have begun to lose their luster. While there are still probably thousands in widespread use throughout the world, evolution has conspired to put them on the endangered species list. Do you remember thicknet, thinnet, or token ring? Yup! That kind of endangered!
Not even 5 years ago, many companies relied on bandwidth from T1, ISDN, or Frame Relay lines to connect to the internet. I still remember trying to get an ISDN line (128k) installed at my house so that I could get high speed internet! A T1 was too costly in those days.
Fast forward to today and we see circuits in the 100M to 10G range at most businesses. Metro Ethernet, DOCSIS, and even DSL provide bandwidth at commodity prices in most urban areas. While there is still a significant footprint of traditional TDM technologies that serve the rural markets, even the $25 per month 3 Mbps DSL availability puts broadband within reach of 95% of businesses.
WAN optimization was in full swing during this limited bandwidth era. It wasn’t economically feasible for a company to install multiple T1s to get the aggregate throughput needed to deliver services. WAN optimizers came to the rescue by using a combination of protocol optimization and data caching, allowing the first computer to download a file, caching the object and chunks of its data, which allowed a subsequent download by other users to be much faster. However, their greatest weaknesses were the UDP protocol and SSL/TLS-encrypted traffic. Applications such as video streaming, or VoIP received no benefit from most optimization solutions. And while these systems can generally accelerate encrypted applications, dealing with key and certificate acquisition and loading created a series of administrative headaches that were so difficult to overcome that most choose to not use this capability.
Jump forward to today’s networks and you begin to see that the cloud and SaaS applications further negate the premise of WAN optimizers. Protocols in use by these services are now optimized for high performance over even slow WAN links. In addition, virtually all of this traffic is SSL/TLS encrypted from the workstation to the cloud using keys that aren’t readily accessible. This means for the WAN optimizer to even see the traffic, you must bring it into the certification path to allow for decryption and re-encryption before delivery. This adds a considerable amount of latency that can cause poor application performance, not to mention that the installation of the certificates and management of the solution is time consuming and cumbersome.
Given the availability of low-cost, high-bandwidth links, and, the pervasive use of SSL/TLS with cloud applications, SD-WAN has taken over as the new standard for performance and reliability in modern networks. The ability to define business parameters and policies for individual applications cannot be matched by any other technology including WAN optimization. It’s inherent ability to constantly health check applications and delivery paths provides for a dynamic ecosystem that greatly improves VoIP, streaming, SaaS, and cloud application performance. It also introduces an additional layer of protection and availability via the use of multiple connections to deliver a complete solution. Highly sensitive traffic such as voice is delivered over the link with the best performance characteristics. Metrics such as jitter, packet loss, and latency are constantly monitored to ensure the best performance.
In today’s cloud-first, high bandwidth world, we are on the cusp of considering WAN optimization to be extinct.
Original Article: Network World