From the University of Nagoya, in Negoya, Japan comes an interesting experiment that NewScientist released to video: a simulation of a traffic jam.

The University simply told 22 drivers to cruise around at a constant 30 kmph, in a circle. At first, the cars were both equidistant and moving constantly. They found, however, that little variations eventually caused the traffic to jam up and created a shockwave of slow moving vehicles that moved backwards along the circle at around 20kmph – similar to traffic jams that are observed in real life.

The experiment has often been simulated in computer models but never run full-scale. What’s interesting is that if they had robots running the experiment, the jam would not be able to occur – it takes human error to cause the fluctuations in behavior necessary to trigger the shockwave jam.

Network traffic follows different rules than motor traffic; but again, human error can lead to “shockwave jamming” as well – which is why you’d find that the most critical traffic is often the most “robot-controlled.” Human beings do not choose the bitrate of most VoIP conversations, for example. Web browsing, on the other hand, has human beings “choosing” how much bandwidth to take – whether to watch the high-def YouTube video, the standard-def YouTube video, or to not watch the YouTube video at all.

This is not to say that misconfiguration issues can’t cause problems, but to suggest that problems with network congestion might be caused by a single, human, user. Again – this is why many companies choose to sequester the “humans” away from the business critical stuff.

Pirate Bay has decided to shut down it’s tracker server, preferring instead to use “trackerless” BitTorrent formats.  Like the move from the centralized Napster to peer-to-peer solutions that didn’t require centralized tracking servers, the Pirate Bay now just becomes a search engine for torrents, not a host for torrent indexes.  It just goes to show you, nothing is more clever than a community of hackers who all missed the previous week’s episode of House.  We may have more on DHT (trackerless torrenting) in a future post, but for now, the basic rundown on how DHT affects network performance can be found at torrentfreak.com

Elsewhere, Mark Cuban has come up with a plan to kill Google.  Literally pay off the top search results for the top keywords to add a robots.txt file to drop out from Google.  Microsoft was willing to pay $45B for Yahoo, and a large number of Web sites with high page rank would opt for, say, a $100M payout to add robots.txt to the file. 

Of course, considering that Google owns a lot of those top sites… YouTube, Blogger.com, it’s a hard road.  The question is: Would most sites risk losing traffic from Google’s results (and their revenue stream) for a quick, one-time payout?  And would people then decide to stick with Google and just not visit those top sites, or move to Yahoo or Bing?  (And what’s stopping Google from retaliating by refusing Yahoo and Bing to index YouTube?) 

One of the big problems with trying to communicate with people outside of IT is that – well, people have misconceptions about what computers are and what you can do with them. From the technical support people who are supposed to “see your screen” to the idea that computers can read your thoughts… well, to many people, computers are like magic to them, so they endow computers with magical properties.

It’s kind of a modern folklore, if you will, not too dissimilar from when people believed that witches’ hexes made cows sick, or that sneezing let demons into your body and caused sickness and madness. It’s a sad but true fact of human nature that we’re more likely to believe superstition than to admit ignorance.

But it doesn’t help when popular culture presents some of these “technological superstitions” as fact.

Recently, there’s a YouTube video going around – a clip from the CBS TV show “Numb3rs,” talking about IRC. It gets the name right – “Internet Relay Chat,” but everything else…

I recently created a video which shows the clip and explains why they get it so wrong. To you and I, this may seem like “explaining the joke,” and in doing so, losing the humor. But to someone who is unfamiliar with IRC or the Internet, it shows why you shouldn’t take your technology cues from so-called “technology”-themed crime dramas.

Just going to write up something quick because I was spending most of the day out with the rest of the marketing dept. working on something fun, but I wanted to draw your attention to today’s strip of “Real Life,” a comic written by gaming geek (and trained chef) Greg Dean, the premise, of course, being that Greg gets to (cheer/cry/laugh/fume) about events in his real life.

I’ve been a reader of “Real Life” since the “Fuego” comic. (I named a tabletop RPG character "Fuego" after the comic. Fuego had no fire-based powers, but he chose the name because "it just sounds so cool.")

I’m copying today’s cartoon for you, and it explains why “fast” Internet isn’t always as “fast” as you’d expect, and it shows the difference between throughput and latency.

Sergey Claus and his helpers, (known collectively as Google Inc.), are planning to give out a few presents to travelers until after the new year – free WiFi at 47 airports. 

Which has me wondering – why isn’t WiFi free at all airports, all the time?  I understand that there’s certainly profit to be made if net addicts or people who simply don’t want to read on a plane pay $10 or $15 to access the Internet for a few hours on grueling layovers, but technology-wise, this seems a no-brainer.  Many airlines offer free wi-fi for business class travelers in their lounges, after all, and if Virgin can offer WiFi in flight, I think the technology has certainly arrived to handle people in a concentrated location on the ground.

Which caused me to wonder – what ever happened to all those plans to cover downtown areas in municipal WiFi?  Oh – that’s right.  It wasn’t profitable enough, so the private companies paid with public funds just decided not to roll them out.  Houston actually took the penalty paid by Earthlink for backing out on the contract to set up a fully public WiFi system in Houston’s poorer neighborhoods. 

What we have here is an “electric car” problem.  We have the technology, we have the ability, the engineering isn’t new, and there’s demand that can be met.  But because the status quo is more profitable with the current technology, the status quo is retained until something comes along which is completely disruptive.

One may think that such a disruption may come from the cellphone market; where 3G and 4G networks provide a promise of Internet interconnectivity anywhere from iPhones and Blackberries.  But even in this market, there are limitations.  Apple iPhones (connected to the AT&T network) have no data caps but cannot tether to laptops, and Verizon’s Droid, which promises tethering in the near future, has a very low data cap – 10GB for $60.  

And all of it comes about because it’s more important for private companies to maximize profit than to maximize technological saturation.  The winning solution isn’t always the superior technology.  You just don’t mind it so much, because most people aren’t aware of what could be. Again – an electric car problem.  We’ve had electric cars for over a decade now and most Americans don’t even know they ever existed – or were driven on Californian roads.  Similarly, people just accept the way things are at airports and in downtown areas, never wondering if things could be better.

The typical way to push superior technology into entrenched markets are either public investment in infrastructure or increasing competition and opening new markets; though I would imagine the barriers of entry into the ISP market are rather high.

But what we can learn from this is that we can – and indeed – have a responsibility to wonder.

Can things be made better?  Is the way we do things the best way.  Obviously, you need objective measurement but ideas should be tested by experiment and service levels need to be verified.  Because you never know.  Maybe there’s a better way. 

Cisco released a slew of collaboration products this week. First, of course, there’s Cisco’s WebEx Mail, a replacement for MS Exchange that integrates with Outlook. There’s the Enterprise Collaboration Platform, essentially a social-networking/company portal kind of thingy that competes with MS Sharepoint. And Enterprise Collaboration Platform is an alternative to IBM’s Lotus Connections.

There are some improvements on competitor’s offerings as well; for example, Cisco includes a service called “pulse” which analyzes social network data as it occurs, and tries to “learn the social graph” associated with corporate networks. And another service, called Show and Share, is sort of like an “internal YouTube” for companies. You can host videos internally and show them to anyone inside the company – but they’re not accessible from the outside – or hosted out on the public Internet. This is also integrated with recording solutions from telepresence to the “widdle iddy biddy” Flip Digital cameras.

It’s a crowded market – not only is Microsoft the established champion, but the scrappy Google has entered in the market as well, with Google Apps for Your Domain. The two approaches to unseat the Microsoft Exchange behemoth are very different, as Google’s solutions require moving things out to the cloud.

For various reasons, companies are often hesitant to trust Google – or any company – with sensitive information like business plans, payroll statistics, personal communications, and the incriminating video of last year’s Christmas Party that you really know you should delete but it’s just too good. There’s also the fact that publically traded companies have to have on-site backups of communications for governmental regulatory reasons anyway, so why not just host the whole thing onsite and eliminate the risk, if it’ll cost about the same anyway?

Cisco’s solutions also integrate with unified communications tools like voice, video, and instant messaging, meaning that the 10 minute phonecall at 10:00 a.m. can be on the network by 10:15 a.m. (Finally, communication tools that travel faster than gossip!)

What impact will these tools have on the network? It’s hard to say, really, depending on each particular enterprise. If companies just switch out one collaboration tool for another, it’s unlikely that there will be much of an increase in traffic. But in other scenarios, it’s difficult to suggest whether overall traffic levels will go up or down. For example, instead of e-mailing large video or audio files back and forth, a unified communications portal allows for a much more simplified “multicast” solution which would take up less bandwidth overall. Then again, the ease of multimedia communication might invite more people to create multimedia communications – thus increasing the amount of bandwidth taken.

One force or the other could dominate, or the two could cancel each other out. In either case, it continues to make a compelling case for maintaining monitoring solutions so that you can catch changes in traffic as they occur, and predict future needs.

Jon on Tech writes about how “The Cloud” is a crock of—well, actually, I’m specifically prohibited from using that kind of language in this blog.

And, to a certain limited extent, he’s right.  “Cloud computing” has been defined and redefined to be almost meaningless.  Anything you can refer to as “the cloud” also has other names elsewhere.  SAAS, Social Networking, Storage – they’re all just big data centers with virtualization.  Skype, OpenID, Google Maps mashups – that’s just all distributed computing. 

“Shock! Horror! It’s starting to sound like The Cloud is just another word for The Internet”

Well, yeah.

Cloud computing is taking hardware and applications which used to run on internal networks and accessing them via the public Internet.  So, in many ways, cloud computing is the Internet.  I don’t think it would be wrong if you exchanged the term “Internet computing” for “cloud computing” in every instance.

So, why do we still use the term “Cloud Computing?”

I think it has to do more with psychology than with technology.  Yes, “Cloud” is a buzzword, but it’s a buzzword that arose in competition with another buzzword: “Internet.”

And that is, there are a lot of people in position to make technical strategy decisions who are afraid of technology – or at least, afraid of what technology means for their business.  For most of the late 1990s and early 2000, the Internet was, if anything, more of a harm to the business than anything else.  First, there was the lost productivity of people surfing on the Internet when they should be working.  And the rise of the Internet also saw the rise in malware, malicious hacking, and the expensive computer security measures designed to counter it.  To paraphrase Dr. McCoy: “[The Internet] is disease and danger wrapped in darkness and silence.”

So the idea that someone would possibly put out their company confidential information, their business contacts, and their mission critical applications out on the Internet sends shivers up the spines of people grown accustomed to associating the Internet with unnecessary risk.  And cat pictures.

So another name was established to explain the benefits of putting business needs out on the Internet.  Something nice and fluffy.  “The Cloud.”  Let’s face it, when people use Facebook or Twitter at home, they don’t think of it as “cloud computing.”  They just view it as “The Internet.”  “The Cloud,” if you’re going to define it, is the word that business prefers to call the Internet. 

It’s a way of making the unfamiliar, familiar.  By calling the Internet “the Cloud” in a business environment, one immediately knows that one is talking about business when using those terms.  It’s the same reason that “money we owe other people” is “Accounts Payable,” and “money other people owe us” is “Accounts Receivable.” 

Not that this has made the problems go away.  One of the reasons it’s called the cloud is because of the lack of visibility and control – an amorphous nature that is vapor-thin.  One in which visibility is difficult, if not impossible – at least, for the most part, until monitoring tools for the cloud are fully developed and implemented. 

You could talk a little bit about “public cloud” versus “private cloud,” the latter term more often associated with virtualized servers in a datacenter environment – but even then you’re talking about concepts akin to “company Intranet.”  Perhaps you’re bringing in a model of service “chargebacks” for computing resources, even internally (with costs on paper) but that model was around since the mainframe era.

So, we’re probably going to continue using the term “The Cloud” when “The Internet” would be just as accurate.  It’s the term that business seems to prefer, and if it means making a few people more comfortable in this strange new century, it’s worth it. 

We’ve been giving out a bunch of advice on how to present information to C-level executives; the overriding points we’ve been hammering are:

• Know your audience and give them summarized, not simplified, information.


• Do not be afraid to make educated guesses when explaining trends, as overall meaning is more important to the C-level executive than pinpoint accuracy.


• Give CIOs and CEOs actionable items when you report – don’t just give them the information without telling them what they can do with that information.

Recently, AskSlashdot fielded a question about how to talk to executives, and in the comments, there’s some good advice and some bad advice on how to do so.

Good Idea - Smooth Wombat:

How about asking them what they want to see? Prepare a short document listing what information you can provide them and in what format, and ask them what they want to see?

Bad Idea – Hognoxious:

Stick a few Trekkish terms in there - quasitron throughput and such - and see if anyone bites. They won’t.

Good Idea – Penguinisto:

Have a very good handle on what you have, and can and cannot be done. Also, never, EVER promise anything up-front without studying the problem.

Bad Idea – Duradin:

The upshot is once you get that report all nice and automated they'll ask you for the exact same report three months later having entirely forgotten its existence. Don't tell them they've been getting that report daily/weekly already for the last three months. They don't like that for some reason. Re-title it, move some columns around, maybe add a new bit of information and then call it good.

Good Idea – Jschen:

Numbers and stats are nice and all, but beyond the headline numbers, your job is to give an executive summary. Here is what I've been doing: “These things are working well. These are improvements that I am targeting or hope to target. Here are the unique challenges and risks that we face and how I plan to deal with them.”

Bad Idea – Yttrstein:

It's always heartwarming to see a Sysadmin ascend to the point where they begin to slowly realize that justification for their salary is going to have to involve some lying.

And finally, I’m not sure if this is a REALLY bad idea, or a REALLY good one from “140Mandak262Jamuna.” I also wonder if this has already been tried…

Collect the amount of water pumped reported by each sensor as a trace between 9:30 AM and 4PM on the days the market is open. Find the correlation between this trace and the S&P500 index with a two minute time lag. See which sensor has a correlation coefficient more than 0.05. Use that info to come up with a trading strategy to buy and sell the exchange traded fund IVV. Propose a project find the leading indicator sensor for more securities like QQQQ, Diamond, XLF, XLU, XLV, XLP and the stock ANSS. …. Build an empire under you. Watch the cash flow of the company. Just before it goes bust, put all this experience in a resume and get a job in the ultra high speed trading division of Morgan Stanley.

I have a new respect for the guy that has to decide QoS policies for the enterprise. Not that I didn’t before, but recently I had to start worrying about QoS policies at home.

Allow me to explain. Earlier this year, Time Warner Cable decided it was going to roll out caps on broadband in Austin, much as it had done in Beaumont, Texas. Doing a quick calculation, I estimated that those caps were going to cost me around $300 a month. Since I didn’t have any other broadband providers in the complex I was currently living in, I decided to move back in with my old roommates, in a 4/4 condo. This condo came with free internet – Time Warner’s business class, which was not going to be capped. Even if it was, the condo also had access to AT&T DSL, so we could always switch if Time Warner brought in caps.

The bad news is that the free connection on the Internet was limited to 2mbps down, 256kbps up, and because of the middleware involved by the condo owner, there were frequent connection problems and just plain connection destructive behavior. Uploading to YouTube often resulted in failed file transfers due to the low speeds. Upgrading the connection through the condo was possible, but the prices, based on Time Warner’s business class prices, were exorbitant. The only upside was that each port had it’s own dedicated 2mbps connection, so whatever happened on one person’s computer didn’t affect anyone else.

Time Warner was expected to roll out broadband caps in September. September has come and gone, and it looks like Time Warner has abandoned plans for broadband caps in Austin after the glum response they’ve gotten from the high-tech test markets. (Looks like I didn’t have to move after all, though my roommates are still pretty cool. ) So, with the understanding we’d switch to AT&T or go back to the condo’s broadband if Time Warner ever did roll-out caps, we decided to get TW residential service in the condo – and since there were four of us sharing the connection, we decided to get the highest speed service they offered – 22mbit down, 2mbit up. (The actual speed was closer to 18mbit down, 1.6mbit up, but still impressive.)

Thing is, for the first time, my three roommates and I were on a shared Internet connection. And we’re all geeks of high magnitude. I have a media center PC, desktop for video editing, laptop for travel, and an iPhone, Patrick, Brendon, and Mike all have laptops, and Mike has an Xbox 360.

It actually went pretty smoothly, considering the relative complexity of the network. The Xbox and the Media Center both get wired connections, everything else runs on wireless. Simple enough, and we were up and running within hours.

It wasn’t until the next day when I noticed something strange. I would queue up a whole bunch of downloads on the Media Center PC, and it would saturate the connection. Doing some Web browsing from my computer in my room however, was slow. YouTube videos stuttered. Congestion had reared its ugly head.

But, hey, I thought to myself. I work for a company that deals with this sort of stuff all the time. QoS policies are the solution! And Tomato Firmware had support for QoS policies, so this was a cinch! I’d be ready to go in no time.

Didn’t quite work out that way. Yes, I could set QoS policies, and I made a reasonable guess as to what those policies should be, but I realized, with some dawning sense of irony, that I had no idea what to give high priority, and what to give low priority. I made some guesses, of course, retaining the defaults of giving high priority to the first 512k of packets on Port 80, but lower priority to anything past 512k (likely a file transfer). Then I gave the Media Center low priority, because, by definition, anything done on that box was entertainment, not work or communication. And Skype, Team Fortress 2, World of Warcraft, and the Xbox Live all got the highest priority – all were time sensitive applications. But after that, I drew a blank. Quite simply, I realized I was making guesses. What was the name of that new game that Mike bought? Borderlands or something? What port does that run on? How do I set high priority for Netflix streaming, and only Netflix streaming? What about YouTube? And how much uploading do we do, anyway?

Quite simply, I had no visibility into my network. Tomato Router is great, but it’s not going to give you information like what applications are running on the network, or what sites are drawing the most traffic during what hours. And even though I had power – I didn’t have the knowledge to make the most effective use of it.

I’m sure many in IT management have gone through something similar – ending up with lopsided traffic policies where 10-20 apps ended up in the highest priority class, because they didn’t have a historical analysis of their network to reference.

All the stuff I’ve been preaching over the last three years – they had finally hit me at home. That’s a sobering thought. And this is just one night. We have no way – other than to try to watch a YouTube video in our rooms while downloading off the Media PC – to determine if we see improvement.

I’m going to need to sit my roommates down soon and talk to them about what gets highest priority. This type of discussion really should occur before deployment, but I just didn’t think it through.

At least I can upload YouTube videos now.

This post was compiled with the help of Jesse Najera, Technical Marketing Engineer