This Industry Viewpoint was contributed by David W Wang
The SD-WAN (software defined – wide area network) era has come. A few years ago, buzzwords like SDN (software defined network) and NFV (network function virtualization) were mostly network architectural concepts, but now SD-WAN has evolved into a disruptive technology and service that can truly revolutionize and benefit the IT/Telecom market and industry on a global scale.
Fundamentally, a SD-WAN, like a virtual PC that separates its operating system from the hardware shell, separates Layer 3 network control from Layer 2 data link. As the result, a lot of network management, optimization and upgrading functionalities that used to be handled by CPE (customer premise equipment) or network edge routers or switches, now can be virtually centralized and managed from the core network facilities via software applications.
The benefits of SD-WAN are surely game changing and disruptive in comparison with the legacy WAN technologies like MPLS (Multiprotocol Label Switching) in terms of network cost, agility, provisioning lead time, management, security, and so on. For instance, SD-WAN will permit users to supplement or replace their dedicated private MPLS with public broadband connectivity, with little sacrificing on the connection reliability, latency control, network security, and bandwidth availability.
Major US telecom carriers are now all pushing for massive SD-WAN deployment to replace the legacy network architecture and functionalities. Recently Verizon has announced that it plans to deliver virtual CPE services to over 30 global markets by the end of the year, allowing business customers to take advantage of software-based services powered by the telco’s cloud based wired and wireless networks. Considering CPE as being pivotal to managed WAN services but for years having been troublesome in sourcing, delivering and provisioning in the global market for Verizon, now this “virtual CPE service” should mark a milestone leap for the telco to boost its WAN solutions beyond the legacy MPLS products.
AT&T is more aggressive in rolling out SD-WAN solutions. Recently the telco has officially launched its NFV on demand service in 76 countries and territories, allowing multinational businesses to reduce network costs and service activation time. The service can cut hardware requirements, so businesses will have much less to manage on site with more cost saving. Meanwhile it simplifies the process of purchasing and adding new network functions from months and weeks to just a few minutes.
As of October 2015, CenturyLink has equipped over 36 locations in multiple geographies globally with its VNS (virtual network services). Then the telco’s latest service, CenturyLink SD-WAN, is an offering that bundles connectivity from several partners, CPE, software licensing, and configuration and monitoring, all accessible from an online service portal.
With all these SD-WAN deployment and launching campaigns underway, the next mission for the service providers will be about how to effectively promote and sell SD-WAN to the marketplace. Whenever a new technology emerges, there will be some early adopters, but carrying on the new service quickly into the main stream may become quite some challenge. While SD-WAN offers quite attractive savings, features and lead-time, some enterprises might remain skeptical on factors like standards and technology maturity, integration with existing architecture, reliability SLA, traffic security, and control power on the network.
In order to speed up the market adoption of SD-WAN, while in the back office the service providers should continue to work closely with equipment vendors to upscale and standardize the new SD-WAN architecture and ecosystem, the front office, on the other hand, may start target marketing and selling SD-WAN to enterprises with the following profiles:
First, existing heavy but unhappy MPLS users. For those enterprises (e.g. banks, retailers, and manufacturers) with over 30 MPLS sites currently, especially when those sites are made of many branch offices across large geographic areas, maintaining, trouble shooting and upgrading this MPLS network can become quite a pain in terms of cost, resources and time. Thus SD-WAN offers may become an easy sale pitch to these enterprise users. The sound approach is to identify the pain points with the current MPLS and then offer SD-WAN as a cure, or at least a partial cure like a hybrid solution to replace some of the MPLS connections.
Second, those enterprises that generate heavy video traffic and on a CDN (content delivery network) platform. Video content includes both regular recreational stream (e.g. from YouTube and online gaming) and business critical purpose like video conferencing, and they take a lot of bandwidth away in network transport. Hence it is beneficial to distinguish them and keep the business critical only video traffic over MPLS, and allocate and reroute those recreational video traffic to SD-WAN, which can lead to better cost and bandwidth control to the enterprise clients.
Third, the SMB sector. Some misunderstanding is SD-WAN is only suited for large enterprises. As a matter of fact, SD-WAN will fit well to those SMBs who want to link their branch offices and remote working employees via an Intranet but cannot afford MPLS or have issues with legacy IP VPN services. In such scenarios, the SD-WAN package can include both public Internet and 4G LTE wireless connections, because for those customers who feel a wireline Internet connection not feasible to pursue, then a 4G LTE solution with up to 300Mb/s download and 75Mb/s upload speed can be the right choice.
Forth, those enterprises that have largely adopted cloud IaaS (infrastructure as a service) and SaaS (software as a service). For these cloud know how customers, since they’ve taken the initial step to virtualize their IT operations and architecture, now it should make a lot of sense for them to virtualize and simplify their network connectivity as well, based on the same cloud centric technologies towards a new ecosystem. Currently they may use a private and public Internet hybrid to handle their cloud communications, but tomorrow they will be better off to use a new hybrid: private MPLS for mission critical data, and SD-WAN for regular data, which is more cost effective, managed and secured.
Finally, those IoT (Internet of things) and M2M (machine to machine) service providers and users. These clients have been struggling between a private MPLS and public Internet. The former can be too much for IoT and M2M applications, while the latter can be vulnerable and uncertain in reliability and security. So now SD-WAN can come up as the best solution to handle IoT and M2M traffic over the IP network globally.
Besides, service providers should offer more educational and training programs, free trial, use cases and testimonials, etc to better promote and sell SD-WAN into the market. If we say 2016 is the year of SD-WAN deployment from the service providers’ standpoint, then 2017 will become the year of SD-WAN adoption for the enterprise market, in particular for those enterprises as profiled and targeted above.
If you haven't already, please take our Reader Survey! Just 3 questions to help us better understand who is reading Telecom Ramblings so we can serve you better!
Categories: Industry Viewpoint · NFV · SDN
Join Our LinkedIn Group
There is too much confusion around SDWAN just like there was around “Cloud” a few years ago. let’s agree on a couple of initial items:
1. Using a portal or simple software as an interface to equipment is not a new technology.
2. Mapping traffic base on the application is not new (Cos/QoS). Many firewalls have been capable of deep packet inspection (Layer 7) for many years now.
3. Application Performance Management or APM has been around for over a decade. We called them Probes. And some simply used IPSLA.
3. If you must acquire new hardware to secure “SDWAN” services aren’t you just buying another router with firewall-like capabilities (Layer 7 visibility)?
I encourage everyone to simply read and understand what you are being offered by the Service Provider or SDWAN hardware provider. Chances are, the Service provider has selected and possibly white-labeled an SDWAN provider’s box and their Dashboard.
Kudos to Meraki. The first to leverage their centralized management portal to control all AP’s via one pane of glass. I’d throw that in the SDN group for the fact that a centralized Software was managing all endpoints.
But what about the WAN in SDWAN?
If I am merely selecting a path A, B, or C (MPLS, Internet, LTE, etc.) is that WAN? Or, am I a box merely deciding on which existing and “Paid for” WAN path to traverse?
Is that SDWAN? That’s like someone telling you which lane cars should go in, and then touting that they build the interstate for you. MPLS, VPLS, and Private Lines still work. They just cost too much and don’t have the flexibility and scale that we need these days.
But if I use the Internet as my WAN preference, I have to worry about Security and Performance. If I can connect Palo Alto to Chicago securely and fully encrypted over the Internet, I can control prioritization of traffic flows, but not the Internet congestion, latency, performance, or path of the Internet (Interstate).
QoS or SDWAN Application Prioritization techniques are only needed/used during instances of congestion. Does anyone want to run a network that is constantly congested? If you are sizing your WAN that tightly, you’re asking for problems. QoS should be treated as an insurance policy. Do we buy car insurance so that we can drive recklessly because we’re covered? Probably not.
MPLS and other WAN topologies are not dead. The way they are delivered today is dying. Folks are flocking to SDWAN providers because it gives them visibility on traffic that they never had or were unable to take the time to institute management software. It allows them to look at a friendlier GUI/Dashboard that is less confusing than their command line interface (CLI).
Think about it now. If buying a tertiary box that helps you decide which of your existing WAN links you should use and when, then buy away.
If you no longer care about your routing protocols running on your WAN and think that complete replacement of your MPLS to and Internet-based WAN with zero control of your Internet speeds and paths, then buy away.
I encourage you all. Be smart. Ask the hard questions. Is SDWAN merely a pig with Lipstick? We’ve had QoS, Layer 7 application control, and dashboards for years… Make no mistake, there are significant cost savings to be realized. We have some great platforms out there that offer intuitive centralization. Just don’t be fooled.
I truly believe that REAL SDWAN is just that: Software-based, controlled and defined Wide Area Networking Services.
Wes Jensen
Co-Founder & CEO
Netrolix
I agree by surface, a lot of network management and features are already available one way or another. But now SD-WAN integrates them into a new ecosystem. Another good analogy is think the legacy IP VPN as a freshman while new SD-WAN as a senior scholar, because SD-WAS has higher level of intelligence. Using network congestion and latency issues over public Internet as an example, now SD-WAN can optimize the routes and quickly re-route and backup the traffic via the mesh network. That’s my 2 cents.
Like I said, really do your homework. 95% of SDWAN providers do not support dynamic routing protocols. Their flows are often static site to site. In addition, the failover/load balancing is done via static configurations and Anycast load balancing. This basically means that there is no failover but rather a duplicate stream of data across multiple paths. When SDWAN providers show demos where they pull the plug and the call stays up, that is because they were streaming the call on both routes MPLS and Internet, or for that matter, any two access options. So, if your goals is to go to SDWAN to save on bandwidth, this seems pretty counterintuitive. Like I said be smart! Ask the right questions. know what you are getting. Hope that helps.
Good points in general, except the MPLS and Internet backup hybrid part should be normal and are not consuming bandwidth in parallel, rather it backs up each other with one as the primary pipe and the other as a secondary idle.
Working for a major telco’s NOC, I used to manually set all these up like: Dynamic Multipoint VPN, Cisco Performance Routing (PfR), and real-time quality measurements, the resulting configuration is complex.
A real SD-WAN is supposed to handle this work for an organization in an automated way, routing and rerouting traffic dynamically based on the current state of the network. The IT team tells the SD-WAN application how certain traffic should be treated, and the solution takes care of the rest. To be more precise, the complexity doesn’t actually go away — it’s simply hidden by the SD-WAN application doing all of the heavy lifting.
Agreed on the hidden SDWAN application comment. but there are capabilities and functionality that is lost. Just know what you’re buying. As an example, would you replace your MPLS network with NGNAT or site to site IPSec?
No need to replace MPLS, instead a hybrid solution makes the best of both MPLS and SD-WAN. Btw, site to site IPsec is an old timer, hardware centric solution. SD-WAN will replace it.
“Wes Is Awesome”
You Are! Thanks!
For the most part, SDWAN is just CCNA in a box. It makes a bunch of functions that a CCNA (or higher) would have to figure out.
You keep saying that word. I do not think SDWAN means what you think it means.
At Airespring we looked at 32 SDWAN vendors, selected 2 so far, 3rd maybe in the mix. We aggregate 18+ carriers to our national datacenters so customers have the widest choice. The right SDWAN solution corrects for lost packets, jitter, out of order packets, asymmetric bandwidth and a lot more while supporting robust routing protocols (bgp, ospf, etc), application QoE/QoS via identity engine that have 2500-10,000 apps already built-in, sub-second failover, AES256-bit encryption, dynamic full mesh or hub/spoke or hybrid topologies, flexible licensing, virtual or hw form factor, cloud connectivity, wan optimization (vendor specific), integrated firewalls or service stacking/chaining, and much more. Link to me on linked in or email me mike.chase@airespring.com but SDWAN is real; it’ll eventually morph into universal CPE (see AT&T flexware and others emerging). -Mike Chase, CCIE# 7226