Newsletter Sign-Up FaceBook LinkedIn Twitter Subscribe INTEROP LOCATIONS:   LAS VEGAS  •   TOKYO  •   MUMBAI  •   NEW YORK

Archive for the 'Las Vegas 2007' Category

May 4th, 2007 | Brian Chee

Can your network really take the load?

Brian Chee

Interop Tested:
Some of the Ixia PR folks brought up a very real example of just how bad things can get if you ignore stress testing your network before you throw it to the wolves. Gee, I truly wonder if our friends at Research in Motion (RIM) invested in traffic simulations before they went live with that last big upgrade? At Interop we don’t leave things like this to chance and this year we’re working with Ixia Communications out of Calabassas, California (aka Los Angeles) to confirm that the InteropNET is ready for prime time.

Checkout Ixia in booth #1024

Imagine my pleasure when I found that one of my ex-students was on the Ixia team assigned to stress test the InteropNET. His job was to test various aspects of the InteropNET under real world traffic simulations. Not just running simulated layer 3 traffic, but getting all the way up into layer 7 (web, video, dhcp, dns, etc) to make sure the entire system was ready to the potential load that both vendors and attendees will generate.

Daniel Bui of Ixia Communications:
As parts of the Interop network came online during hotstage, Ixia has been testing the performance in the areas of the Extreme Show Floor Backbone, DNS server, DHCP server, Nortel off Show Backbone and Xirrus Wireless Access Points. We started with a simple L3 throughput test between PEDs 101 - Ped 108 (show floor equipment racks) using IxNetwork. Our tests confirmed that the Backbone is able to handle Attendee and booth traffic.

Once we established that our L2 and L3 tests ran smoothly, we then move up the OSI stack and tested the Core, Off show floor and off show floor by simulating web, upload and download transactions using IxLoad. We then started to test the DNS and DHCP server by sending patterns of DNS queries and DHCP requests, ramped up to and past expected production levels. Finally, we simulated web transactions using the Application test in IxLoad from the show floor to the external link. In this case the target was some Ixia gear in the Qwest colocation facility in San Jose. The Application test records the user experience by capturing the activities of the user and measuring the delay time between activities and responses. These are called think times. So we simulate this by going out to real websites like Google, Yahoo, CNN etc. So while we wanted to stress test everything, we cranked everything back for a softer touch on the public websites.

Since the InteropNET will also be running both Video Conferencing (Life Size HD Video Conferencing) and streaming video of keynotes; it was key to make sure that we tested for key things like: latency, jitter, packet fragmentation and a few others to confirm that Real Time Streaming Protocol (RTSP) and Video on Demand (VoD) would be rock solid.

Since so many of the Interop Attendess will be depending upon wireless to get email, we placed IxWLAN (WiFi ) testers in PEDs where we anticipate heavy usage of the Xirrus wireless access points. IxWLAN was then used to create simulated users and real world traffic (such as HTTP, FTP, and RTSP) to confirm that the Xirrus can handle the traffic expected at the show.

The reality is that too few IT upgrades are stress tested before deployment, even when they can affect mission critical operations. The Ixia suite of traffic simulation systems go far beyond just dumping traffic onto the network, it generates intelligent traffic patterns to simulate what you might actually see. Don’t just cross your fingers, Ixia test it and like the InteropNET Team, sleep better at night knowing that upgrade or deployment isn’t going to melt at the first sign of a traffic spike.

Brian Chee is a researcher at the University of Hawaii School of Ocean and Earth Sciences and Technology (SOEST) and runs the Advanced Network Computing Laboratory (ANCL). He is also a Senior Contributing Editor at InfoWorld Magazine with ANCL playing host to some of InfoWorld’s big iron reviews.

May 1st, 2007 | Brian Chee

Avaya and DiVitas mobility integration demo

Brian Chee

At the 2007 Las Vegas Interop HotStage (where we build the InteropNET first in a warehouse in Northern California) Avaya and DiVitas demonstrated a cellular-wifi-wired VoIP integration without the need for placement of gear in the telephone company central office.

I got the chance to look at the DiVitas solution almost a year ago for InfoWorld and have seen it grow and become increasingly polished solution to help enterprises get better control over their cellular minutes. While this may sound like a bummer for the cellular carriers, it isn’t. This solution is giving enterprise IT more control over an expense wildcard and for the first time is opening doors for the cellular carriers into a previously sticky market. I for one have had massive roaming bills when doing projects overseas, explaning my agency’s reluctance for providing cellular technology to anyone other than top management. This type of control and integration into the enterprise telephony system can keep costs under control while giving the road warrior equal access to enterprise resources.

This short video (just under 5 minutes of flash video) is a live demonstration of the DiVitas Mobility system roaming with a dual mode phone (cellular and WiFi) from WiFi to the Avaya deskset and back. The typical scenario would be a call started while driving, transition to the enterprise WiFi (in our case the Xirrus WiFi system) and then be able to pickup the same call on the desk telephone. You can also extend the call back to the cellular phone if you need to leave your desk.

Click here to go to the InfoWorld site and view the video…


Brian Chee is a researcher at the University of Hawaii School of Ocean and Earth Sciences and Technology (SOEST) and runs the Advanced Network Computing Laboratory (ANCL). He is also a Senior Contributing Editor at InfoWorld Magazine with ANCL playing host to some of InfoWorld’s big iron reviews.

Apr 24th, 2007 | Brian Chee

802.1x Authentication, but wait…oops!

Brian Chee

****Warning: Propeller head stuff****

So here we are cleaning up the warehouse getting the network pretty for press night at HotStage. Just about everything is working and now folks are busy typing away trying to document everything.

However, creating screen shots for user documentation means we need to recreate the flow of events that user would go through and in the case of using the Microsoft 802.1x supplicant (think logging into a switch port to confirm that you’re a valid user…protects from bad people plugging into the NOC network and running amok.) once you’ve logged in and the system has installed the authentication values, it doesn’t normally reprompt you to present login credentials. The bummer here is that our 802.1x NOC person (Lisa Lorenzin of Juniper Networks) started asking around whether anyone knew of an elegant way to force the Microsoft supplicant to ignore the old credentials and re-prompt for the new set. The answer came as a team effort with folks from Extreme Networks, the iLabs engineers and this reporter (observering by hovering over shoulders) got to the mysterious registry key.

So if you pop open this graphic, you’ll find one occurrence of the magic registry key that you delete.

So to get here I did a search in regedit for “eapol” (you may need to do this multiple times.)

The magic action is to delete “UserEapInfo” and then disable that network interface and then re-enable it. Viola, now Windows acts like it has never seen that authenticated network before, and it will re-prompt you for your credentials.

So much thanks to Mike Scripek of Extreme Networks, Craig Watkin of the iLabs, Lisa Lorenzin of Juniper for a truly useful tidbit of information. Why you might ask? Simple, NAC is coming due to all the bad kiddies out in the world forcing enterprise network managers to distrust strangers. So as more and more enterprises start down the road to NAC, be sure to take note of this little registry key.


Brian Chee is a researcher at the University of Hawaii School of Ocean and Earth Sciences and Technology (SOEST) and runs the Advanced Network Computing Laboratory (ANCL). He is also a Senior Contributing Editor at InfoWorld Magazine with ANCL playing host to some of InfoWorld’s big iron reviews.

Apr 24th, 2007 | Brian Chee

One ringie dingie….two ringie dingie…

Brian Chee

We’ve come a long way from the Lily Tomlinson telephone operator skit and so has the mobile phone. One of the things that Interop has been exceptional at is tossing vendors together in an environment that encourages everyone to play nice. This year we’re experimenting with a combination of DiVitas Networks and Avaya to bring voice connectivity to the NOC team wherever they may be.

Here’s Edwin of DiVitas with his collection of test phones. What makes this all so very cool is that they make use of a dual mode phone (WiFi and Cellular) to let the NOC team take advantage of the WiFi all through the Mandalay Bay Convention Center. When a call comes in for a DiVitas Mobility Client the system determines whether the call quality is better through the cellular carrier or via the WiFi. So in all reality as long as the team members are inside the convention center we shouldn’t be burning any cellular minutes. What’s even more cool is that through some interoperability testing with Avaya (Thank you Paul Humphries!) we can start a call on cellular, walk into the convention center and once the system confirms a quality WiFi connection (it doesn’t flip immediately to avoid thrashing), it will switch. Better yet, we can also have the call ring the Avaya deskphone and be able to simply pickup the desk phone to already find the call waiting.

Now that’s all fine and good, but almost all of the dual mode phones out in the world are 802.11g only, and with the sheer number of WiFi demonstrations that will be running in the convention center it will almost certainly turn the 802.11b/g (2.4ghz uniband) range into WiFi soup. (Hint, go out into the outer areas of the convention center to avoid your WiFi connection from dropping so often) To solve this problem we’re also going to be using 802.11a based VoIP phones from Spectra Link.

While these don’t have cellular radios the SpectraLink phones will give the NOC team voice connectivity away from the deskphones and in some cases may replace the desk phone.

At the heart of our voice system is the Avaya Communications Manager 4.0 with Avaya’s SES (SIP Enablement Services) where the Avaya PBX can provide some Avaya specific advanced services while still providing connectivity options for 3rd party SIP equipment and services like those provided by DiVitas.

So while Avaya and DiVitas don’t normally play together, the flexibility of the Avaya Communications Manager and the work done by the IETF on the SIP standards all make for INTEROPerability.


Brian Chee is a researcher at the University of Hawaii School of Ocean and Earth Sciences and Technology (SOEST) and runs the Advanced Network Computing Laboratory (ANCL). He is also a Senior Contributing Editor at InfoWorld Magazine with ANCL playing host to some of InfoWorld’s big iron reviews.

Apr 23rd, 2007 | Brian Chee

How you keep everyone on the same page…

Brian Chee

I love the concept of remote collaboration, but let’s face it. When I’m part of a WebEx or a Microsoft LiveMeeting and connect to a meeting room, the folks a the corporate office always use the white board. Then they try desparately to describe when they’re drawing on the board…my knee jerk reaction is to then mute my phone and go to sleep. I know others are in the same boat when I got exactly the same reaction from some Microsoft engineers in my lab the other week. So let’s just face up to reality, a picture is worth a million words, so how about sharing that picture with the rest of us…

Virtual Ink (now owned by Rubbermaid Newell) has been around for quite a while, but for some reason hasn’t been able to break out of the education market. This portable device is even better than sliced bread, and is now one of my travel companions at Interop. Here at HotStage, the engineers tend to redesign at the drop of a hat, and even with most of us in a single location, it’s still not easy to stay on the same page. However, now when we hide in the conference room, I drag along my laptop and my wireless Mimio Xi so that every stroke on the white board is saved and then distributed to everyone in the meeting.

So while this may look like chicken scratches in the dirt, it’s diagrams like this that go into the building of the InteropNET. Our fearless leader (Glenn Evans) scribbles his way across the board, while the rest of us roll our eyes back in our heads when we realize just how much work this means.

Well the reality is that it could be more if we didn’t have the same notes. Just keep in mind the if you ask 10 people to tell you about a bank robbery, you’ll get 10 different versions. Not something you can afford in a network expected to deliver 100% uptime for the class rooms, meeting rooms, exhibit floor, wireless, Voice over IP Telephone System, etc, etc… So while this may sound like a gimmick, just think of this product the next time you’re on a conference call and someone starts in on a white board you can’t see.

Check the Mimio website for compatibility statements for your remote collaboration system, and if you’re looking at using for a conference room in your office, you may also want to check out Wall Talker. A wall covering that doubles as both a dry erase marker surface and a projection surface that’s a perfect match for the Mimio stick. Just keep in mind that the maximum size the Mimio can support is 4 feet by 8 feet. With Mimio studio and a projector, you could project a world map on the board, markup it up and save both as a single image.

Here’s a scanned brochure from www.walltalkers.com


Brian Chee is a researcher at the University of Hawaii School of Ocean and Earth Sciences and Technology (SOEST) and runs the Advanced Network Computing Laboratory (ANCL). He is also a Senior Contributing Editor with InfoWorld Magazine with ANCL playing host to some of InfoWorld’s big iron reviews.

Apr 21st, 2007 | Brian Chee

BgInfo v4.10 from Microsoft

Brian Chee

So you vendors out there…as you get your booth demos ready for the show floor, you might want to consider adding this little accessory to the demo machines to make it easier to keep them all apart.

I totally agree with the article written by Bryce Cogswell Published: March 16, 2007 in the Microsoft TechNet. When you have a bunch of machines all configured similarly for either testing, or just a rack of similar machines doing different things. The desktops all start to look the same, especially if you’re buring the midnight oil on deadline. It’s just all to easy to make a modification to the wrong machine.

Our of self defense I started editing the wall paper with the machine names in my lab, but Microsoft has released a really cool tool that does such things automatically. As an added bonus, if you put it into the scheduler, it will refresh the image periodically with current information on critical items like the amount of free space left. The amount and type of information is configurable during setup.

Neal Allen of Fluke Networks brought this to my attention (thank you Neal) at Interop HotStage. His luggable had it running, and he’s setup it up on the desktops of the virtual machines he’s got running on the lunchbox luggable the Fluke folk use for testing. Wow, so simple, but yet oh so very useful.

Kudos to the Microsoft folks for a wall paper that totally makes sense.

*NOTE: much thanks to Microsoft for NOT locking down their images on the technet site.

/brian chee


Brian Chee is a researcher at the University of Hawaii School of Ocean and Earth Sciences and Technology (SOEST) and runs the Advanced Network Computing Laboratory (ANCL). He is also a Senior Contributing Editor with InfoWorld Magazine with ANCL playing host to some of InfoWorld’s big iron reviews.

Apr 20th, 2007 | Robert Ballecer, SJ

Interop Rising

Originally posted on TheTechStop.net

The Interop crew is will into setup at our super-secret facilities in California. It’s late (or early) and I’m tired, so I’m just going to post a few pictures that will hopefully give you a teaser of what you can expect on the Interop.net blog as well as on TheTechStop.net

Stay Tuned!

Apr 17th, 2007 | Brian Chee

If we build it, they will come?

Brian Chee

I first experienced the InteropNET at the Las Vegas 1995 show and haven’t wanted to stop since. It must have been in the mid 1970’s that I got some advice from then Lt. Cmr Gordon Eubanks (He’s gone on to bigger things since) while guest lecturing at Chaminade University. The best engineers aren’t the ones that know it all, but rather the ones that know who to call. So while it might have been Dr. Norman Abramson (Mr. AlohaNET) the point is that this industry learns from each other and working on the InteropNET never fails to illustrate that no matter how much you know, someone else knows more.

This extremely talented group of volunteer engineers twice annually will gather in a warehouse in Belmont, California (San Mateo area) to rack, stack, configure and document the world’s largest temporary network. Key to the sucess is a time tested design that provides for the InteropNET Team to quickly deploy a network similar in size to many large corporate networks in just a few days. So as the pieces arrive on our loading docks, I hear the chatter of engineers designing their prospective areas of responsibility.

So as we gain momentum, look for more and more postings as we piece together this huge jigsaw puzzle we call the InteropNET.

We don’t just build it and cross our fingers. This year Ixia corporation is providing a collection of test equipment and engineering that allow us to run simulated traffic across the InteropNET. This warm and fuzzy gives us the confidence to handle the inrush of booth networks that make up the exhibit floor and meeting rooms.

We organize the racks by function…here’s an engineer from Extreme Networks in front of our two Core Racks…we actually call them PEDS (left over terminolog from when Interop was held in Monterey and had all the gear stacked on bar tables or pedestals) You can clearly see the Extreme purple, the Juniper Blue+White, and the GigaMON orange…

So while we do have an out of band access system (we call it Access Ether) the initial configs sometimes are done from box tops. Here’s Pat D. of Network General working on his InfiniStream data capture boxes.


Brian Chee is a researcher at the University of Hawaii School of Ocean and Earth Sciences and Technology (SOEST) and runs the Advanced Network Computing Laboratory (ANCL). He is also a Senior Contributing Editor with InfoWorld Magazine with ANCL playing host to some of InfoWorld’s big iron reviews.

Mar 19th, 2007 | Robert Ballecer, SJ

Infrant NV+ — Big things in Small Packages

Full article @ TheTechStop.net

Interop Las Vegas 2007 is just over two months away, so I guess it’s time to start gathering all our techno-eggs into the basket that we call “Hot Stage.” For two week, the engineers of Interop will descend upon the Interop warehouse and piece together the most advanced temporary network in the world. With millions of dollars of equipment being generously supplied by our Vendor-Sponsors, everything from the latest in wireless to the fastest in switching, Hot Stage is somewhat of an uberGeek Christmas.

Still… sometimes even uberGeeks need a little bling.

TheTechStop.net received a NV+ from Infrant Technologies to evaluate as a potential “Support Solution” for Interop — basically, we wanted to bring in gear that could show off the amazing capabilities of the InteropNet — We were so impressed by the stability, security and performance of the box, that we featured it in an episode of gadget, an IPTV show that caters to uberGeeks.

If you’re looking for high-performance RAID storage for your SMB, you need to take a look at the NV+ — check out the episode and decide for yourself if its what you need for your network.

Feb 13th, 2007 | Thomas Maufer

Impossible to spell interoperability without Interop

I have been associated with the InteropNet since before the WWW hit in 1994. Yes, Virginia, there was an Internet before the WWW. :-) The thing that made the InteropNet valuable from the earliest days (when it was simply a TCP/IP plugfest) was that the InteropNet enabled multiple vendors to prove interoperability for multiple protocols at the same time in the same place using a stable network infrastructure: the InteropNet.

But what is ‘interoperability’?

The most famous aphorism about interoperability is Postel’s Law: Be conservative in what you send, and be liberal in what you receive. Both parts are equally important. In fact, as is noted in section 4.5 of RFC 3117, too much emphasis on the “liberal in what you receive” part, without enough focus on the “conservative in what you send part” part, is going to lead to progressively poorer quality implementations of protocols.

One reason for that effect is that people can only judge their implementations against other specific implementations…not against all possible implementations. One thing that can help avoid the problems with implementations being too lax in their RX processing, while not strict enough in their TX processing, is a thorough exposure to a wide variety of invalid input. One way to do this is through protocol “fuzzing” — but random input is too simplistic: it is not repeatable, and a QA engineer or developer also needs to expose their product to other real implementations (as is done in the context of the InteropNet) and to a testbed that supports regression across time.

A usable protocol torture test is everything that the Internet is *not* — the Internet can reorder, corrupt, drop, modify, truncate, and do other kinds of unintentional mischief. Moreover, these abused packets don’t just come from your own implementation of a given protocol…other implementations that are not based on the same reading of the spec also need to be processed. Finally, on top of all that, specific intentional malicious traffic (from hackers) is going to arrive at your doorstep. All of these effects are going to challenge the best implementation, but they are not repeatable, and any bug reports from such effects are going to be virtually impossible to reproduce. The most important attribute of a productive protocol torture test is that it be part of a repeatable test harness.

The InteropNet constrains the test space to at least provide a stable environment in which to conduct interoperability testing. Our customers are actively using Security Analyzers to challenge many of the protocol implementations in the InteropNet during hot staging so that the robustness can be enhanced prior to the deployment in the live shows in Las Vegas and New York. (Disclaimer: I work for Mu Security; we provide the InteropNet’s Security Analyzers.)

The goal is to raise the security and robustness bar on protocol implementation quality across as many InteropNet devices as possible — to eliminate the background noise (from the “obvious” flaws in RX processing). The result is improved productivity of interoperability testing on the InteropNet — the biggest “win” is in improving interoperability by having implementations converge ever more closely to the specs, being as strict as possible on the TX traffic, while having progressively more robust error-handling on RX that can tolerate a wide variety of invalid input (without ecouraging laziness on the part of implementers). An important side effect of improved interoperability is that more robust implementations of protocols are less vulnerable to attack by hackers.

Next »