By Art Reisman

The other day, I sold a smart refrigerator  to a customer. When they found out it had a computer in it, and could be controlled remotely from the Internet, they asked me if they could run it on their Virtual Machine to save some space in the kitchen.  I told them, sure  we support that , they just need to get a-hold of  an  add-on compressor and a 40 foot cubic container module for their VM,  and we would just ship a plug-in application. There would be no need to ship any hardware to them, we have  a virtual refrigerator!

I purposely used that over the top analogy, to highlight,  the chill down my spine I feel, when I hear about vendors bundling their core network equipment into a VM.

Virtual machines make a lot of sense for somebody running a data center with 10 different servers and consolidating them into one box.   My underlying discomfort stems from the extension of  that mission onto equipment that is involved the real-time transport of your data.  Switches , routers , firewalls and bandwidth shapers.  Why do I feel this way? Am I just an old stubborn  engineer clinging to the old ways while the world passed me by?

Not really, we have set up virtual machines with our bandwidth shaper with success in our labs, it is actually pretty cool. My discomfort arises with the fact that bandwidth shapers are finely tuned, real-time devices, with software that must run at the core level of the computer’s operating system.  A bandwidth shaper must have absolute control of perhaps 4 ethernet/fiber ports or more and under no circumstances can it compete with  CPU resources should a server become overloaded.  The consequences of any resource contention are at best a slow internet, and at worst a complete lock up.   Yes I understand a in theory a modern VM can divvy up resources , but how do we ensure that it is done correctly ?   When we ship a standalone device running only our application we know  exactly what it is capable of,  and since we have thousands of identical configuration in the field,  we know that the technology configuration that leaves our factory dock is rock solid stable.

This is not to say we will never offer a virtual machine, we did have one customer where the logistics of their set up was so remote that the benefits of a virtual bandwidth  shaper on their standard configuration far outweighed the risks I mentioned above; but for the most part saving a few dollars on rack space and an extra piece of hardware are not worth the jeopardizing the stability of a critical piece of in-line equipment.

By Art Reisman

CTO http://www.apconnections.net

Below are my predictions for technology in 2018 and beyond. As you will see some of them are fairly pragmatic, while others may stretch the imagination a little bit.

1. Forget about drones delivering packages to your door; too many obstacles in densely populated areas. For example, I don’t want unmanned drones dangling 30 pound flower pots flying above my head in my neighborhood. One gust of wind and bam,  flower-pot comes hurtling out of the sky.  I don’t want it even if it is technically possible!  But what is feasible, and likely, are slow plodding autonomous robots that can carry a payload and navigate to your doorstep.   Not as sexy as zippy little drones, but this technology is fairly mature on factory floors already, and those robots don’t ask for much in return.
2. As for Networking advancements, we may see a “Cloud” backlash where companies bring some of their technology back in-house to gain full control of their systems and data.  I am not predicting the Cloud won’t continue to be a big player, it will, but it may have a hiccup or two along the way.  My reasoning is simple, and it goes back to the days of the telephone when AT&T started offering a PBX in the sky.  The exact name for this service slips my mind.  It sounded great and had its advantages, but many companies opted to purchase their own customer premise PBX equipment, as they did not want a third-party operating such a critical piece of infrastructure.  The same might be said for private companies thinking about the Cloud.  They could make an argument that they need to secure their own data and also ensure uptime access to their data.
3. More broadband wireless ISPs coming to your neighborhood as an alternative option for home Internet.  I have had my ear to the street for quite some time, and the ability to beam high-speed Internet to your house has come a long way in the last 10 years.  Also the distrust, bitterness, dare I say hatred, for the traditional large incumbents is always a factor. One friend of mine is making inroads in a major city right in the heart of downtown simply by word of mouth.  His speeds are competitive, his costs are lower, and his service cannot be matched by the entrenched incumbent.
4. Lower automobile insurance rates.  The newer fleet of smart cars that automatically break for or completely avoid obstacles is going to reduce serious accidents by 50 percent or more in the near future.  Insurance payouts will drop and eventually this will be passed on to consumers.  Longer-term, as everyone on the road has autonomous driving cars, insurance will be analogous to a manufacturer’s warranty, and will be paid by the auto manufacturer.
5. The Internet of Things (IoT) will continue to explode, particularly in the smart home arena.  Home security has taken leaps & bounds in recent years, enabling a consumer to lock/unlock, view and manage their home remotely.  Now we are seeing IoT imbedded in more appliances, which will be able to be controlled remotely as well – so that you can run the dishwasher, washer, dryer, or oven from anywhere.
6. Individual Biosensory data, like that collected by Garmin and Fitbit monitors, will be used by more companies and in more ways.  In 2018 my health insurance company is offering discounts for members that prove they use their gym memberships.  It is only a small leap to imagine a health insurance company asking for my biosensory data, to select my insurance group and to set my insurance rates.  As more people use fitness trackers and share their data (currently only with friends), it will become the norm to share this type of data, probably at first anonymously.  I can see a future where  health care providers and employers use this data to make decisions.

I will update soon as new ideas continue to pop into my head all the time.  Stay tuned!

In a previous article we made the following statement.

“ISPs are now promising 100 megabit per second consumer  service , and are betting on the fact that most consumers will only use a fraction of that at any given time.  In other words, they have oversold their capacity without backlash.  In the unlikely event that all their customers tried to pull their max bandwidth at one time, there would be extreme gridlock, but the probability of this happening is almost zero. “

So I ask the question  , what would it take to make full use of your 100 megabit pipe ?

A typical  streamed movie consumes about 4 megabits, So you would need to watch 25 Netflix movies at once all day every day to fully utilize your pipe.  Obviously watching 25 movies at once all day every day is not very practical, you’d need multiple Netflix Accounts and 25 devices to watch them on.

Big files:  A 100 Gigabyte file, that’s a good size download for a consumer right?   Well, that would take approximately 4 minutes  to download on a 100 megabit pipe, and then you’d have to find another one.

For convenience  maybe you could find   a 1,000 Gigabyte file? That would take  only 40 minutes, so you are still kind of left with a good deal of spare pipe  for most of the day.  How about a 10,000 Gigabyte file  (10 Tera Bytes ) , that would take 400 minutes.   By my calculations, in order to make use of  your 100 megabit  pipe completely for 24 hours , you would need to download a 40 Terabyte file?

Where could you find such a file?

I did some poking around and there are a couple of sites that have gigantic files for no particular reason , but the only practical file with a reason to download  was this one:

“Some time ago I was interested in creating custom maps of the Earth, and I realized that the data files needed for this are pretty large; and the more zoom you want, the larger are the data files.

OpenStreetMap has a huge file of the Earth which is 82GB compressed and around 1TB uncompressed according to the OSM wiki, and it will become larger. You can find it updated here.”

So this very large file that maps the entire earth is 82 Gig in compressed form for download, a tiny fraction of the full 40 terabytes  you would need to download in one day to fill up your pipe.

What is the moral of the story ?

Internet providers can safely offer 100 megabit pipes full well knowing,  that  even their heaviest users are likely not going to average  more than 5 megabits  sustained over a long time period.  You would actually have to be maliciously downloading ridiculously sized files all day every day to use your full pipe.

By Art Reisman

CTO http://www.netequalizer.com

I have  a confession to make. Even though we have a corporate e-mail server at my company, I have been using Gmail for my primary business e-mail going back to 2002.  I  love the ability to search old records and conversations from the past.   With Google’s technology , searching gmail was second to none. Sometimes , I searched just for nostalgia  purposes, like the final e-mail conversation I had wih my Mom when my dad was taken off dialysis in hospice, and sometimes for business reasons.   Unfortunately my world has recently been shattered.   All my e-mail prior to 2008 is completely gone, and I have searched far and wide for a policy from Gmail that might explain why.  I pay a monthly fee for google storage and was well below my limit, I have tried their support forums and so far  just silence. If you are a long time e-mail users I suggest you try to search  your archives.  Ten years seems to be the cut off where things got dumped or lost.

And no ,  have not been corresponding with any Russian operatives!