For the last several columns we've looked at the key factors affecting WAN performance – loss, latency, jitter and bandwidth – and various ways of addressing these issues. As 2013 begins, I'd like to step back and look at the bigger picture, before any further dives down into the details.
A long-time networking guy, I've been passionate about Enterprise WANs for a decade now. I became passionate about the basic idea of this Next-generation Enterprise WAN (NEW) architecture after talking with a prospective customer in 2008 about how he had leveraged server virtualization to consolidate the 40+ different applications his company was using onto four Dell superservers, which he deployed at a colocation facility. That was my "aha!," light-bulb-over-the-head moment.
I'm not sure whether the term "cloud computing" was in vogue yet or not in 2008 (it certainly soon would be), or if people were still using the term "utility computing" for the data center computing portion of this, but being already familiar with what was going on with networking technologies for the enterprise WAN, this was when I first understood the possibilities of, and power behind, combining these technologies together, and how it was going to change enterprise networking even as "cloud computing" was changing how we think about computing. And one of the most exciting parts is that this change in the enterprise WAN would both be fueled by, and help accelerate, the changes going on with cloud computing, and public and hybrid clouds in particular.
I chose the title of this column to help convey my passion on the subject. "Generational change" in enterprise computing or networking typically only happens once every 10 to 15 years. With "next-generation" in the title, I quite deliberately have chosen to focus on those of you innovators, early adopters and the earliest of the early majority who are seeking to gain a competitive advantage for your companies and/or your careers by finding and riding one of the next waves.
We've spent a fair amount of time discussing the five technologies I consider the key components of the NEW architecture: WAN Optimization, server virtualization, distributed/replicated file service, colocation, and WAN Virtualization. And we most certainly will be discussing each of them further, including their roles and their interactions, in the months ahead. The existence of these technologies is, of course, critical to the NEW architecture. In 2008, when the general idea of this first dawned on me, it was still too early for this kind of a sea change in architecture to occur. Almost five years later, the time has arrived. Here, I want to look at the broader networking and computing conditions that make the timing ripe now for this next-generation approach.
Sinek claims that part of the power of "start with why" is based on biology and about appealing to people's intuition. I don't know about the biology part, but let me lay out the things that you probably need to believe if you are to buy in to this generational change I see coming.
Said another way, if you don't believe most of the following things, then you probably either won't like, or won't be particularly interested in, most of what you are likely to see in this column in 2013.
- Multiprotocol Label Switching (MPLS) services might decline in price, but they will never come close to the price/bit delivered by broadband or by colo-based Internet pricing, because the carriers won't crash their own profitable market by slashing prices 90%+.
- Internet pricing (price/bit) will continue to come down roughly with Moore's Law. Internet broadband pricing is already 30-100x better price/bit than MPLS at small sites, and Internet price/bit at colo facilities is already 8-20x better than the price of MPLS at fiber-connected corporate data centers and headquarters locations.
- The consumer market will remain a driver for the carriers to invest in Internet and wireless networks, and these will continue to improve price/performance, and for competitive reasons slowly improve or at worst stay the same in terms of reliability, even though none of these networks on its own will achieve enterprise-quality reliability for those same competitive economic reasons.
- WAN bandwidth needs will grow far faster than private WAN budgets, so ever-larger MPLS pipes won't be the answer.
- Stored video files will become important to many enterprises.
- Videoconferencing is becoming important to many enterprises, but the large bandwidth and reliability requirements don't fit with MPLS costs.
- Colocation facilities will continue to attract vendors of all types, keeping WAN bandwidth costs low and offering an increasing variety of interesting services.
- Public cloud services and SaaS will continue to find a foothold and become an important part of most enterprise IT shops.
- Most larger enterprises won't go for fully distributed Internet access over a single local link for accessing critical cloud-based services.
- ...but backhauling all Internet traffic to headquarters or a corporate data center over thin, expensive MPLS pipes won't scale sufficiently or offer good enough performance access for SaaS or public cloud-based services for globally distributed remote office users.
- Enterprises can't – or won't – count on the individual cloud services providers to deliver performance predictability to all their users at all their locations for their most mission critical applications.
- Few enterprises will go to purely private clouds with no SaaS; fewer still will move completely to the public cloud.
- Larger enterprises will prefer to centralize WAN and security complexity at a handful of locations, in the same way they are consolidating data centers.
- There is now a critical mass of enterprise network managers who are familiar with deploying equipment/software/applications at colocation facilities, thanks to prior web hosting and backup/disaster recovery initiatives.
- As they face pressure from CIOs and line managers to support moves "to the cloud," network managers will be looking for technologies that can deliver the reliability, visibility and control over their WANs that they are used to from their private WANs.
- Managed Services Providers seeking a larger share of the enterprise market will need to solve the WAN reliability challenges of the public Internet if they want to accelerate the trend of outsourcing application services.
If you do believe most or all of the above, then I hope you'll stay on for the Next-generation Enterprise WAN ride in 2013. I think you'll find it worthwhile. Rather, I believe that you will!
A twenty-five year data networking veteran, Andy founded Talari Networks, a pioneer in WAN Virtualization technology, and served as its first CEO, and is now leading product management at Aryaka Networks. Andy is the author of an upcoming book on Next-generation Enterprise WANs.