The view from seat 22A
Long ago in a faraway Internet backbone galaxy
You remember 2005, don’t you? Back then you could book wholesale IP transit for about $15 (€18 in Europe) per megabit a month. Data volume commits were much smaller then, as little as 1Gb. Those were the good old days. Today, traffic volumes have increased by as much as 10 times, yet many providers have seen little growth in total IP transit revenues.
Some claimed Cogent wasn’t so cogent
I remember a European carrier crying in his beer that Cogent was encroaching in its geography, making sales at about half the customary price of €18/M/M. Well, boo hoo. Half of the wholesale IP transit players started gobbling up the other half in 2005: I lost half my potential Market Intelligence sales to that M&A frenzy.
I remember, back in 2006 I got a cool logo baseball cap from the Cogent sales team at a LINX meeting. I rather liked it … for a while. During the following few years I was accosted whenever I wore it. Cogent is doing this. Cogent is doing that. Cogent is destroying the business.
There’s more to the Cogent story than lower prices and increased volumes
True, the Cogent enterprise telesales team was offering everyone with an AS a deal-of-the-month, but their wholesale team was far more savvy — they were offering a “great” deal. Even Renesys was tempted. They made us a retail offer of $50/M/M for a small 10M test application, but if we committed to 100M we’d get $20/M/M! It seemed too good to pass up. But in a fit of common sense we did the math and determined that in the not-so-unlikely event that we did not actually use the 100M, we would end up paying more than the retail market $50/M/M price.
Not surprisingly, a significant number of IP transit customers used Cogent’s low pricing offer to shop around for deals. Oh, and they neglected to mention it was an offer based on a higher commit level and surprisingly the NSPs neglected to check it out. In my opinion, those NSPs who matched the low pricing without doing their homework did just as much as Cogent did to depress IP transit pricing.
Rushing around Russia
The IP transit market is mostly commoditized now, with just a few regions and paths still commanding premium pricing. Asia is generally underserved with capacity; and IP transit suppliers are beginning to get competitive in the Middle East. Russia, however, now offers a reliable low-latency alternative to the Asia-to-Europe undersea cables. So Jim Cowie (our CTO) and I went to Moscow to see what was happening.
Meeting with some of our Russian backbone providers was interesting, of course, but their travel tips for a couple of out-of-towners were really interesting. The first thing we tried was a high-end cab billed directly to our hotel. We slipped into a sleek Mercedes commanded by a steely-eyed Cossack in a trim grey suit. Our driver knew no fear. Obstacles in our path — mere trifles. We were rocketed to our destination with laser-like focus at warp speed. Vladimir Putin could not have expected more.
Jim’s idea of a good time is haggling for the best price, so we tried a rusty, retro Japanese cab driven by a guy from some former Soviet republic who seemed unaware that first gear wasn’t the only gear. Okay, so the price was right, but I wanted to get back to my hotel before the Olympics start in 2014. Next trip, Jim thinks it will be fun to take the train from the airport into Moscow. Sure hate to miss that, but I have to be in Tokyo.
The sky is falling
Falling prices convinced most NSPs it was time to revamp their marketing focus. Every provider started claiming its product was better and worth more than basic IP transit. Some were more successful than others with this tactic. But when large aggressive search/content providers applied pressure, lower prices prevailed. Yes, you can compensate for low prices with volume, but only if you have sufficient capacity. By sufficient I mean huge, YouTube-like (100G) capacity. Not an option for many smaller NSPs.
In 2007, everyone started rolling out value-added services, such as better service level agreements, lower latency for specific routes, and wiggle room on commit volumes. This proved somewhat effective for some providers, but prices continued to erode. There had to be a better way … I know! Akami has good margins! Let’s do what they’re doing! We’ll sell CDN services!
Level 3 quickly acquired the SAVVIS CDN, while many other providers just resold CDNs. Tata for example started selling BitGravity, while Deutsche Telekom sold EdgeCast. That’s when CDN pricing started to resemble IP pricing. It seems that as soon as this slew of new NSP offerings inundated the market, CDN service became a commodity at about twice the speed of underlying IP transit.
For better or worse everyone did not jump into CDN
There are industry gurus who think large-commit CDN deals provide little pricing advantage over basic IP transit. Some NSPs, such as Sprint, refocused their sales activities on enterprise-managed services where pricing and margins proved more stable. This turned out to be a double-edged light saber for Sprint. Level 3 snatched their 1st place position in the Renesys Global Internet Ranking, leaving Sprint to scramble in a dead heat for 2nd place with Global Crossing.
Clouds are moving in
By 2008 (perhaps 06 or 07?), along came cloud computing. Led by Amazon and Google, new services cropped up to help enterprise customers store, manage and deliver their data anywhere they wanted. This represented a whole new market opportunity that enabled companies to outsource the headache of managing their own data. Selling cloud computing services is really hot now and it makes a lot of sense. Network and hosting facilities already exist within the NSPs, and many customers are eager to take advantage of inexpensive cloud resources. But — and it’s a big BUT — there are legitimate security concerns. Saving 20% up front on your IT budget is no savings at all if an insecure cloud allows someone to steal proprietary information that cost millions of dollars to develop. Enter Renesys with innovative cloud mapping solutions to provide enhanced security (more on this soon), my chance to win back business lost to a rapidly consolidating NSP market.
It’s only natural
So, with perfect 20/20 hindsight, we can see that all the hand-wringing about the collapse of the IP transit market was just that — a lot of hand-wringing. Commoditization of markets is a natural, if uncomfortable, progression. Here’s what I predict:
- Continued consolidation of the major NSPs.
- A reevaluation of the business case for extensive peering. If transit is cheap enough, the expense of negotiating and establishing bilateral peering agreements for a small fraction of your traffic might not be worth it.
- The transit-free cartel, Tier-1, will increase to about 20 NSPs over the next few years.
- Sales teams remain focused on new high-margin products based on IP transit.
- Seat 22A won’t cool off anytime soon. I’ll be selling Renesys cloud intelligence solutions from Fargo, North Dakota to Timbuktu.
Cooling my heels in Madrid
The latest acquisition to rock the Internet backbone market is Tinet (globally ranked 8th), gobbled up by wholesale voice solutions provider Neutral Tandem. But the Internet backbone isn’t alone in this M&A mania. Other, seemingly unrelated, industries are also rife with commoditization and consolidation. British Airways, for example, just merged with Iberia.
Bear with me. This matters. I fly a lot. And a lot of it is on OneWorld carriers, so it seemed reasonable to fly Iberia through Madrid on my way back from Moscow. Not!
I had an early flight out of Moscow with a 3-hour layover in Madrid, so I wasn’t too concerned when we left 2 hours late. Okay, maybe I was a little concerned when we landed at Madrid and stopped miles short of the terminal even though there were lots of empty gates. We filed obediently onto buses, which deposited us at the immigration and transit area. All of us lucky US-bound passengers had to go through additional security — full body scans, pat-downs, etc., eventually. It was at least 20 minutes before they got their act together and started up the scanners and baggage X-ray machines.
I love sprinting with carry-on bags. Builds character and aerobic capacity. I arrived at my gate a little short of breath, but not short on time. This plane was late too.
I hope Internet consolidation is a bit better organized. Next time I’m flying through Tokyo.