The Peering Pendulum Swings

Ask Dr. Peering

Q:
DrPeering -

End of the year again... What will we see in the peering ecosystem in 2015?
Chef Stacy
A:
Chef -

EXPECT SIGNIFICANT CHANGES IN THE PEERING WORLD IN 2015

Expect to see the peering world continue its trajectory of evolution, further dissolving peering boundaries. Peering will evolve from single-location/multi-location peering where there is existing critical mass; the paradigm will scale towards a more distributed peering model where anyone can peer in any location from any IXP. The Open-IX initiative proved that the peering systems can change, so peering coordinators continue exploring alternatives blends for optimal traffic exchange.

The ease and speed of provisioning will drive peering beyond the market of traditional ISPs and large content; we will all see a much larger market of small network operators (1-3Gbps) and enterprises (2-200Gbps) who were previously unable to justify peering. Now the enterprise has direct connections to their mission critical network services.

In 2015, the broader market will see the peering secrets revealed:
peering is not difficult,
peering delivers value without having to watch it,
peering pretty much stays up and keeps working., and if it fails, there is always transit to automatically fail back to.

So why not add peering in all of its forms (local, remote, paid, distributed, etc.) to the blend?

The interconnection pendulum has swung from centralized (NSFNET) to carrier-IXPs (post-NSFNET) to distributed (metro-market peering) back to centralized (carrier-neutral) and it is swinging back to distributed.

Worldwide Peering Education Tour
To help communicate this message of how to, and why to peer, DrPeering will continue his evangelism, traveling around the world signing books, training companies about peering, and leading peering tutorials.

Will see you at a peering event in Honolulu in January, San Antonio in Feb, Oklahoma city in late February or March, the Bahamas and San Antonio in April. Please reach out if there is an opportunity to present the peering message.

Q:
DrPeering -

Would peering have helped Sony?

John Holmes
A:

John -

Sony suffered a significant penetration, allowing unauthorized access to privileged information. According to the New York Times, the hackers stole the systems administrator credentials and passwords. Then the contracts, salaries, emails, budgets, medical records, social security numbers, etc. were published and 75% of Sony servers and data were destroyed. No amount of peering would have protected against inherent security vulnerabilities.

What peering can do is architecturally segregate traffic, a common security technique, and peering also establishes a direct escalation path with peered networks to deal with technical issues related to traffic exchange.

I use the word “architecturally” because once established, nothing is actively required to maintain the traffic optimization. If peering goes away, there is always transit to back up to. By the way, large financial institutes like Visa peer because it architecturally improves security.

Traffic segregation is also helpful during times of Denial of Service (DoS) and Distributed Denial of Services (DDoS) attacks. If you contact an ISP Network Operations Center and you aren’t a customer or a peer, they most likely won’t take your call. So we have seen that peering is helpful when coordination is helpful. Some peering coordinators view this coordination feature as the strongest reason to peer.

Traffic segregation can also be helpful for traffic management. Peering allows you to direct and control traffic the way that transit can not. The largest content companies in the world peer to manage and improve the end-user experience. Peering also scales very well - one can aggregate traffic across public peering fabrics and the peaks of one flow can offset the valleys in other flows.

Q:
DrPeering -

Agree with you that cell phones sure do suck.

Here is my test --> listen to hold music on your cell phone. Try it...

Do you hear 1960’s AM radio quality or worse?

I can’t believe we have to suffer through this junky service in 2014. We need a high def phone system that at least returns us to the quality of land lines of the 1970’s.

Thela Hun Gingeet
A:
Thela - we used to use music as a sound source in the early Internet days. We would have audio conferences and sometimes video conference calls, and back then we would spend the first half hour stumbling over the technology issues. When someone was speaking, it was never clear if they paused or if there was sound lost due to a problem. We seem to have accepted this 1990’s era terrible service level as “good enough.”

Music is indeed a good test. You can easily hear when there are artifacts in the codecs or transmission - we all know how a popular song is supposed to sound.

The other anomaly that drives me nuts is the latency. Sometimes it seems like we have to go back to saying “over” after completing a phrase. I really don’t understand how people cut the landline cord.

Peering of course may be part of the solution - directly connecting may remove network hops and decrease the latency slightly, but it is unlikely to correct problems related to the variety of clever and mildly incompatible technologies now in between the phone devices.

A collection of peering prescriptions and predictions for 2015.

The Peering Pendulum Swings

December 31, 2014

Next

The 2014 Internet Peering Playbook

In Print

and for the Kindle:

The PDF, ePub and

.mobi files are

also available as a

perpetually updated

DropBox share:

Price: $9.99

and in French!