As we pass the one-week anniversary of World IPv6 Day, some highlights on our experiences:

• v6 Day brought close to a two-fold increase in IPv6 traffic on the Voxel network, with a lot of that traffic remaining following its conclusion.  At the same time, IPv6 traffic volumes on our network remain extremely low, at well under 1% of our total Internet traffic.
• In the days leading up to the 8th, we sharpened our on-call engineering rotation, as well as briefed all customer-facing staff of the day’s possible implications.  In the end, there were no issues of note; our v6d-related tickets were limited to customers wanting to join in the fun at the last minute (not a problem, thanks to the dual-stacked load balancers we stood up specifically for this), and a Verisign salesman attempting to sell us on Mafia-style DDoS “protection” (ironic in that the Verisign web site was one of few experiencing reachability issues on the 8th).  Not a single non-reachability issue was reported.
• Though there were no non-reachability issues encountered, the day nonetheless provided for an excellent learning experience.  Working off feedback from our customers and providers, we applied various tweaks to our routing policies and provider preferences, along with load balancer configuration defaults, to optimize traffic flow.
• Many participating customers chose to continue publishing DNS AAAA records — effectively making their sites reachable over IPv6 — indefinitely.  Those who didn’t left the door open to future participation in the near-term, pending additional analytics.

These experiences were shared by our colleagues at large web properties also participating, almost identically.  All told, big content brought its “A game” on the 8th, failing to disappoint.

Not to be overshadowed, Internet Society (ISOC) did a great job rallying the troops, coordinating communications between participants, and increasing awareness among the general public.  Even PBS’s Gwen Ifill knew what was happening, thanks to the ISOC’s leadership:

https://www.pbs.org/newshour/bb/science/jan-june11/ipv6_06-08.html

So, where do we go from here?  I’d personally like to see repeat days, perhaps even an “IPv6 Week” or “IPv6 Month”, with increased participation from Internet access providers.

For the most part, here in the United States, it’s impossible to purchase IPv6-enabled Internet access from the major incumbents.  My Verizon Wireless Android phone enjoys dual-stack IPv4/IPv6 handset connectivity when in 4G LTE coverage areas (very cool and progressive in and of itself) — sadly tethering is limited to an IPv4 double NAT (really, guys?) — and that’s about it for options.

Comcast Cable leads the pack with its excellent technology trials and community involvement, however the percentage of actual households utilizing IPv6-capable Internet service is very small; they’re publishing exact statistics soon.  Unfortunately, the followers are few and far between, with no other cable modem providers announcing deployment plans with concrete dates.

Suspiciously absent from meaningful participation on the 8th was Verizon Telecom, who’d be well-served to follow the lead of their colleagues in Wireless.  What they lacked in participation, they made up for with a laundry list of everything that could possibly go wrong, in detail vivid enough to scare off even the largest v6 proponent.  I suppose that’s all one can do when FiOS — a next-generation ETHERNET based service, where you’d expect these things to be easy given a lack of legacy technology, combined with their colossal advantage of controlling the home gateway — has no v6 plans announced for 2011.  AT&T has been similarly quiet, other than to announce that IPv6 is coming to their next-generation UVerse technology over a hodge-podge of tunnels, and not speaking to the other half of their footprint stuck with older DSL.

Why does this all matter?  Here at Voxel, our focus is on ensuring the best possible Internet experience for our customers, both today and in the future.

As we run out of Internet addresses, carriers are preparing to position broadband subscribers behind provider-side (IPv4) NATs, termed “Carrier Grade” NATs, or “CGNs” for short.  I worry these devices might bring with them new performance bottlenecks and troubleshooting complexity for the operators and their users — and by extension, us.  In addition, CGNs will most assuredly break the “end-to-end” reachability required for some customer applications in use today, such as gaming and peer-to-peer streaming.

Equally important is that our business is growing, and we continue to board new customers and applications requiring IP resources.  While we’ll try as hard as we can to always have IPv4 addresses available, even utilizing the secondary market and provider assignments where absolutely necessary, one must consider that the pool will be fully dry at some future point in time.  In a hypothetical world with no available v4 IPs, continuing to service our customers will require additional complexity, something which we’re beginning to model and test out today.  Required will be additional translation technologies hosted on our side, such as load-balancers, NAT64/DNS64, and VPNs to facilitate administrative access.

In sum, IPv6 on both the hosting and access side is a key requirement to ensuring service continuity.  Just as we’ve helped countless customers enable their content and applications for IPv6, we look forward to heightened awareness and adaptation among our colleagues in the access business.

Viddler, a platform for video publishers, wrote on their blog this past week about how they chose Internap to help them with their goal to deliver “greater uptime and service reliability” to their customers. This is a major goal for Viddler in 2011 and we are very proud that Viddler chose Internap to help them accomplish it. We look forward to working with Viddler.

Personally, I am very excited about Viddler choosing Internap for our Internet transit and infrastructure services. I was first introduced to Viddler by Gary Vaynerchuk, a social media pioneer, several years ago and I used Viddler for a business I ran at the time. Viddler made the video uploading and posting process very easy and I considered them my platform provider of choice.

Get familiar with Viddler, read their blog post about why they chose Internap, and enjoy the service as much as I did.

Thanks to Ian Borg and the team at Viddler for the wonderful mention!

Last week, we held our webcast on how to boost the availability and performance of your multi-homed networks using Internap’s Flow Control Platform™ (FCP) intelligent route optimization appliance. As such, we wanted to share the Q&A here. As a reminder, the FCP sits in your multi-homed network environment and performs three essential functions:

• Monitors your networks to identify network issues
• Automatically performs routing changes to route your traffic across the most optimal path to your end-user
• Provides extensive network reporting and simulation functions for a full view into your multi-homed network environments

Questions & Answers (Q&A) from the Webcast

“Does the FCP sit in-line to the network and if so, does this represent a Single Point of Failure (SPOF)?”

FCP does not sit in line of your network traffic. Your traffic flows through the same chain of equipment as it did previously. There is no Single Point of Failure risk to your network.

“What performance improvements have your customers experienced through implementation of FCP?”

Internap has a very broad set of customers using FCP – Financial, Online Trading, Online Retail, Gaming, etc. The feedback we get is that companies see up to a 40% improvement in performance through the reduction of latency they were previously experiencing.

“How does the routing change downstream traffic?”

FCP’s traffic routing is done at the point of egress. FCP looks at end-to-end traffic performance to your destination and the quality of this link to make a determination as to the best carrier and path to route your traffic.

“Does the FCP optimize incoming traffic?”

No. The FCP optimizes your outgoing traffic across the network.

“Do you offer a class of service for different providers?”

FCP does have configuration options that allow companies to route traffic across their chosen carriers based on performance or cost or a combination of the two characteristics [FCP provides the ability to configure policies in its FlowView Manager panel].

Our Performance IP™ service allows customers to leverage our Managed Internet Route Optimizer™ technology to route their traffic across a broader set of carriers that Internap connects to via our Private Network Access Point (P-NAP®)architecture.

“How much capacity can FCP handle?”

Flow Control Platform is a scalable solution. FCP comes in two configurations, the FCP-GigE and FCP-10GigE. The FCP-GigE can support up to four GigE connections and 4Gbps. The FCP-10GigE can support up to two 10GigE connections and up to 20Gbps.

With their companion Flow Control Remote (FCR) devices, the FCR-GigE and FCR-10GigE, the solutions scale in increments up to 4Gbps and 20Gbps, respectively, for each additional FCR unit deployed.

“Can you use FCP with your Performance IP connection?”

Absolutely. You can optimize FCP within an existing Performance IP connection. This is the best of both worlds in that customers can leverage the greater diversity of carriers for traffic routing that Internap is connected to. This is not a requirement for using FCP.

Last week, we gave a glimpse into our thinking for the future of data center design in the webcast, “Scaling Your Data Center Footprint for Higher Density”. Our premise: data centers of the future must shift the current relationship between space and power.

In today’s typical raised floor multi-tenant data centers, the ability to meet tenants’ growing demands for power is met by allocating more and more space or “footprint.” Current cooling limitations mean that power is capped to a maximum 6-8kW per rack enclosure which is insufficient to meet growing tenant needs. As a result, tenants must acquire additional footprint in the facility to accommodate growth, at additional cost.

In next generation multi-tenant data centers, support for much higher rack densities will be critical. Colocation tenants must have the ability to scale on-demand within their existing footprint and not be forced to sprawl across a facility. As a result, the relationship between space and power must be redefined to power and cooling whereby both are available to support much higher density rack configurations.

Companies should carefully consider their current colocation provider’s ability to support their high density rack configurations in the future.

Internap is working to develop flexible colocation solutions that can support higher density rack configurations requiring up to 8kW+ to 18kW+ in power consumption. Doing so will meet the needs of our customers who desire greater longevity within the same facility and an enhanced ROI for their colocated solutions.

We attend a number of industry events each year and pay attention to what attendees are telling us. As its name connotes, IT Roadmap is largely attended by local IT professionals looking to stay abreast of new technologies and practices in the industry. Of course, cloud computing is a key topic and a number of presentations addressed various aspects of cloud computing.

Two observations:

First, IT is still largely in a ‘wait and see mode’ with respect to cloud computing and exploiting cloud’s capabilities. For a variety of reasons, IT hasn’t wholeheartedly embraced public cloud computing in terms of actual adoption. These reasons may be based on security and compliance concerns, application sensitivity concerns, workload (other priorities) concerns, and suitability concerns. It is clear that there is quite a bit of variance between the state of IT in the different companies we heard from and from the timelines these IT organizations expect to deploy cloud solutions more broadly.

In a sense, it appeared that IT is currently absorbed in its own redefinition–to become more meaningful in the eyes of the business– and came looking for information and guidance on how it can continue to evolve. In that respect, the IT professionals didn’t come looking for cloud computing specifically but for perhaps, to learn from their colleagues about how others might be solving similar challenges.

Second, it seemed that even some of the cloud vendors were grappling with where to take the audience with respect to cloud computing. How much did the audience need to be educated? How much emphasis should be placed on public cloud discussion? Was the audience ready for what the presenter was talking to? Did the audience believe their value proposition for cloud computing? Did the material really relate to the real challenges IT in its departments over the next two years?

Summary

Overall, it’s critical from the standpoint of the solution provider to reflect on the state of IT today. In many respects, the way forward for IT and vendors alike at the moment is to agree that that path forward will be more collaborative and evolutionary. This, of course, is difficult for many providers given that adoption of “exciting” new technologies is often in a vacuum from the risk and compliance policies that IT works by.

Providers must provide a broad range of solutions and bring a level of expertise that matches up to what IT really needs right now. We have to remind ourselves that IT needs our help and that the best solution may very well not be the most cutting-edge one. And we can’t place so much emphasis on a single technology to solve IT’s challenges when those challenges cover such a wide spectrum of resource, cost and operational elements. We have to be amenable as much as flexible – IT professionals genuinely need our help.

As a hosting and cloud services provider, we’ve been tooling our systems, network and provisioning and management platforms for IPv6 support years.  Adam Rothschild, our VP of Network Architecture, has been at the forefront of the ISP and network operators communities sharing configurations, developing practical multi-tenant strategies and gathering real-world data. Voxel now proudly supports full “dual-stack” (both IPv4 and IPv6) deployments on all of its hosting services, including our virtual compute service, VoxCLOUD , and our bare metal VoxSERVER platform.

However, what’s really missing in the land of IPv6 right now is real-world deployment practice at service-provider scale.  Hosting companies, backbone providers, CDN’s and content owners/websites all need to work together to make current IPv4 networks and related services actually perform in an IPv6 environment. To that end, Voxel has been an active participant in World IPv6 Day, which is tomorrow, June 8^th, 2011. The goal of this “Test Flight Day” is to motivate organizations across the spectrum to prepare their services for IPv6 and help move along the transition from IPv4 as the address space runs out.

To find out more about World IPv6 Day, please visit the Internet Society website at https://www.worldipv6day.org/.

So What’s With IPv6 on The Cloud?

The explosion in applications, data, traditional IT services and more has only accelerated the exhaustion of IPv4 address space. However, moving an application or service to the cloud is also a great time to plan for IPv6 support. Most end users or application developers adjust or even replace their application when moving from a traditional IT environment to a cloud, where they can take advantage of things like instant provisioning, horizontal scalability, or mulitple geographic locations. Ensuring that your application and underlying cloud provider is ready for IPv6, is critical to any deployment. A few points to check with your provider on are:

• Is their cloud dual stack capable? Can it route both IPv4 and IPv6 addresses?
• Are there any specific limitations when using IPv6 address space, such as tunnels, which can slow down throughput or make your hosts harder to configure?
• Is there a lack of feature parity on such add-ons as load balancing, IP portability, or logging?
• What does the provider’s IPv6 network connectivity look like? Transit and peering will be completely different from an IPv4 network, so it’s key to remember that.
• Are any solutions offered that make transitioning to IPv6 easier?

What Does the IPv6 Internet Look Like Today?

With tomorrow’s World IPv6 Day, we’re looking forward to getting a significant amount of real-world data from our backbone, peering relationships and end-user content websites. We’ll make sure to keep you posted on what worked, what didn’t and how things are looking for an IPv6 only future.