Broadband initiatives - impact to your network?

Mailman List Mirror (Read Only)
1

I'm one of the reporters who covers broadband and cloud computing for InformationWeek magazine (www.informationweek.com), and it's interesting to me that one of the issues with cloud adoption has to do with the limited pipe networks available in this country. For example, it's not feasible to do a massive data load through the networks that are currently available -- you need to FedEx a hard drive to Amazon. Holy cow, it's SneakerNet for the 21st Century!

Initiatives like the federal BTOP (Broadband Technology Opportunity Program) and Google's infamous (and so far invisible) fiber program promise to change, by orders of magnitude, the pipe that's available to most folks, and therefore change equations like SneakerNet for cloud loads. Surely the backbone will also need to be much more capacious, with middle mile changes like this.

I am interested in hearing from folks on this list about the impacts that these programs are having on your network build-out and management plans. I'm also interested in hearing your perspective on real network management issues that could either be caused by or fixed by proposed FCC regulation of broadband.

I don't have a political axe to grind, and I know that broadband is something of a political issue right now. But there are pragmatics that data center managers and CIOs have to deal with every day, and knowing what's coming up on the national broadband agenda is becoming more and more important.

You can reply to me either on-list, or off-list (jf@feldman.org) if you want to remain anonymous. My deadline is Monday, close of business (5pm ET). Thanks so much for your perspective.

--Jonathan

Jonathan Feldman
Contributing Editor, InformationWeek
http://www.informationweek.com
Twitter: @_jfeldman

2

is this a 'this country' bandwidth problem or the problem that moving
10tb of 'corporate data' in a 'secure fashion' from 'office' to
'cloud' really isn't a simple task? and that cutting a DB over at a
point in time 'next tuesday!' is far easier done by shipping a
point-in-time copy of the DB via sata-drive than 'holy cow copy this
over the corp ds3, while we make sure not to kill it for mail/web/etc
other corporate normal uses' ?

The broadband plan stuff mostly covers consumers, not enterprises,
most of the (amazon as the example here) cloud folks offer
disk-delivery options for businesses.

you seem to be comparing apples to oranges, no?

-chris

3

Never underestimate the bandwidth of a stationwagon full of
$current_high_density_storage_media.

Nick

4

More than one person has pointed out that offline media will always be higher bandwidth than transmission lines (but nobody with such elegance and hilarity as Nick Hilliard's last post). Point taken. The question, in my mind, is whether it's reasonable to ask that regional providers reach the same bar as privately owned campus networks.

I don't agree with you, Christopher, that the broadband plan won't affect corporate users. I know that this list _mostly_ consists of operators, but I've gotten some offline responses to my initial query that seem to indicate that enterprise users utilize SOHO (consumer grade, but with higher speeds) for various branch office needs. Also, when a technology gets "consumerized" it tends to create interesting effects in terms of features and price points.

Think of it this way: where would corporate mobile phones be without the consumer effect? We'd still be carrying them around in bags and only corporate officers would have them.

I appreciate everyone's response!

5

... as Andrew T teaches ... :smiley:

6

The question, in my mind, is whether it's reasonable to ask that
regional providers reach the same bar as privately owned campus
networks.

you are comparing LAN to WAN, never a bright idea

randy

7

I've never claimed to be particularly bright, but I do like to challenge assumptions.

I meant "privately owned campuses spanning many miles." Is that a WAN? LAN? "MAN"? Seriously, should there really be a difference? If so, why must there be a difference? Let's not forget that ADSL is distance limited. Should it have ever been classified as a WAN technology? Compare that to fiber-connected Ethernet, a so-called LAN technology that goes miles and miles.

8

It isn't only the amount of bandwidth available but also in many cases
the protocols used to transmit the data. It takes smarter than the
average bear to figure out how to get data across a fat pipe over a long
distance at a high rate. TCP protocols are limited by the number of
packets allowed to be "in flight" according to how the stack is
configured. One might need to go to unorthodox or rather new methods to
use all the available bandwidth.

There are many cases of someone being stymied as to why they can't even
get anywhere near 10 megabits of throughput on a GigE path from Los
Angeles to London using FTP, for example. In many cases the
responsibility of getting data from point A to point B is handled by
people who don't bring their network operators into the discussion where
problems like this can be pointed out to them. Often the first time the
enterprise network group hears about it is when someone complains that
the "fast pipe" to $continent is "slow" and therefore must be broken and
that is generally followed by the demand that it be fixed immediately if
that demand is not included in the first email.

That is when conversations bearing sounds like mpscp and uftp begin and
then someone says "aw, screw it, just send them a disk".

George

9

That is when conversations bearing sounds like mpscp and uftp begin and
then someone says "aw, screw it, just send them a disk".

LOL!!!!

10

I don't agree with you, Christopher, that the broadband plan won't affect
corporate users. I know that this list _mostly_ consists of operators, but

(there are a fair number of consumer network operations folks on nanog
as well...)

There have been plans to offer 'business' connectivity (replacing
T1/T3 last-mile type things) from the likes of Verizon (FiOS) for some
time. To date you can't (and they don't seem to have plans really) get
a last-mile tail on FiOS with BGP for routing information (like for a
redundant connection setup, or for alternate provider paths: FiOS
50mbps link from VZ + 45mbps Ds3 from ATT using BGP to manage your
redundancy needs). I don't know that you could not do the same on
Comcast or Cox's deployments at this time, maybe someone from these
alternatives have already spoken up privately on the matter.

I've gotten some offline responses to my initial query that seem to indicate
that enterprise users utilize SOHO (consumer grade, but with higher speeds)

Sure, lots of folks use 'consumer grade' links for out-sites, that
dish on top of the Mobil station being the cannonical example. These
out-sites don't generally have the data concentration of the main
office, nor the bandwidth needs, nor the redundancy/resiliency needs.

Using a SOHO/Consumer link in the right place is a fine solution,
using it at your core site, not so fine...

for various branch office needs. Also, when a technology gets
"consumerized" it tends to create interesting effects in terms of features
and price points.

Still waiting for that on the FiOS space or the Comcast space (where's
my 100mbps cable/FiOS link with BGP for redundancy?).

I CAN get a 50mbps bidirectional FiOS link with static ip addresses
(that I have to pay for the 'privilege' of having) but I can NOT use
my own ip space, nor can I use a routing protocol to tell VZ or the
rest of the world to prefer my alternate link to get to my office.
That's suboptimal, and not 'business class' service.

Think of it this way: where would corporate mobile phones be without the
consumer effect? We'd still be carrying them around in bags and only
corporate officers would have them.

I'm not sure that the corporate smartphone usage was driven by
consumers, it seems (to me) to be the other way around actually... I'm
not a mobile-maven so who knows :slight_smile:

-Chris

11

I wrote a first round BTOP application.

No, the program doesn't quite promise to change, by orders of magnitude, the pipe that's available to most folks, and even if it did, that isn't a very strong promise.

"Most folks" live in urban areas, adequately served by physics, if not the private, and the surviving public infrastructure. "Most folks" who reside in BTOP eligible area codes are not adequately served by physics, and BTOP is, IMHO, limited solutions to the physics problem, with possibly sustainable public incentive funding.

The "orders of magnitude" claim, and the plural in "orders" is key, is both over blown and misses what is, IMHO, the most interesting aspect of revisiting the physics assumptions about the edge of service. Is unidirectional transport (monitized video streams) the rural service most absent and most valued, or are other characteristics of networks competitive with, or superior to, that service model?

The sneaker net meme is worth holding on to, among others. Some of this was grist for the PILC WG.

I went with Plan B, but then again, my application got zero funding, and folks that follow this may appreciate the relevance of the mapping portion of the BTOP/BIP package to selection, and the role of state government in selection.

I suggest coverage of the lobbying of BTOP/BIP grants is at least as interesting as the problems various applicants attempt to state and provide solutions for.

Held until after 5pm PDT, mostly so I could take a walk.

Eric

12

is geoff's isp business 101 still the canonic reference for what this
reporter needs for clue? doing it micro-incrementally on list is a
major ton of <bleep>.

randy

13

If you drive around rural central and northeastern Texas, every ranch house and bunkhouse has a DirecTV or Dish installation. Surprisingly, many of these same houses also have DSL available from the (heavily subsidized) telephone coops in the area. The speeds aren't screaming, typically being in the 300-700 down/128-384 up ADSL-2+ range. So the demand is there, and so is the service in some areas.

--Chris

14

Jonathan Feldman wrote:

I'm one of the reporters who covers broadband and cloud computing for InformationWeek magazine (www.informationweek.com), and it's interesting to me that one of the issues with cloud adoption has to do with the limited pipe networks available in this country. For example, it's not feasible to do a massive data load through the networks that are currently available -- you need to FedEx a hard drive to Amazon. Holy cow, it's SneakerNet for the 21st Century!

What's wrong with this? It's not feasible to build a network that spans many ISPs and backbones, capable of doing massive data loads, if the demand for these loads (e.g. "upload all our data to a cloud computing system") is infrequent and usually one-time-only - which it seems to be. It's not as if there's a huge performance hit to using FedEx to solve this problem - what is the benefit to the customer in having it all happen within hours instead of 1-2 days?

There are other, far more often desired or accessed services (e.g. video on demand, video teleconferencing) that absolutely need high performance big pipe bandwidth, whose needs can not be met with FedEx. Customers who need to access or offer video-on-demand are far more willing to pay, month after month, for access to a high performance backbone. Your average corporate customer isn't going to be willing to pay month-after-month for a super big super fast pipe (faster than they need for their everyday internet access purposes) just so that they can - once - upload their entire corporate database to "the cloud" faster than they can FedEx disks to their chosen cloud provider.

Look at the business case (or lack thereof) for the service before you ask "why isn't this available". Unless/until there's a business case for many customers to pay for the service, there's not going to be any purpose in creating the product.

jc

15

If the data you need to preload is sufficiently large (e.g. 10s or hundreds of terabytes then yeah it should come as no surprise that it might be more convenient to move by shifting around disks. 100TB of raw disk is around $8000.

16

If the data you need to preload is sufficiently large (e.g. 10s or hundreds of terabytes then yeah it should come as no surprise that it might be more convenient to move by shifting around disks. 100TB of raw disk is around $8000.

The cost of equipment is not the driver here, as you can presumably reuse it.

Looking around, I can find a 2 Terabyte drive with a ship weight of 2 pounds. To ship this from Virginia
to Cupertino, California overnight by FedEx is $ 53.46, and I can mail them back for $ 14.50. Assuming that "overnight" is a 24 hour delay, this is an effective bandwidth of 185 Mbps. If I do this every weekday for a month (20 days), I have shipped 40 Terabytes for $ 1359.20, so I have an effective "work week bandwidth cost" of $ 7.34 / Mbps / Month, which seems fairly competitive, especially as I can turn this on and off as needed (e.g., I don't have to pay for Holidays).

So, depending on need, the shipment of physical media may be cost competitive, as well as merely convenient.

Regards
Marshall

17

Even ATM years ago blurred that arbitrary line.

Why does there even need to be a line between local and wide in terms of networking? As far as IP is concerned, there is no difference. Even as far as Ethernet is concerned, there is no difference. It's ATM's promise all over again with people reinventing wheels that shouldn't have to be reinvented....WAN's exist for demarcation, typically, at least in the way I've used them (I used a POS OC3 over a 35 mile path for three years as a LANish link, with the WAN link that had BGP speakers attached being FastEthernet... talk about blurring a line; and now I use a L3VPN tunnel on a WANish Metro Ethernet link to replace the direct OC3 LAN link....))

The BTOP applications of which I'm familiar could just as easily carry traffic that would traditionally be classified as local area; or even storage area, for that matter, as fibre channel in particular does very well over long distances (run IP on FC, and get better than Ethernet throughput for less money, even.... :-)). Drop a wave mux in, hit it with intermediate reach optics, up the number of buffers (on FC, at least, and pay the license for the larger buffers, to vendor B at least), and drop your storage elsewhere. I'd rather get a wave than IP or even SONET transport any day. Wish it were an option here.

And I see the BTOP ARRA apps having the potential, if done right, to extend the 'LAN' (as opposed to 'broadcast domain' even though I know many use the two terms synonymously) to a global reach.

WAN's historically have been differentiated by lower bandwidth, greater segmentation/demarcation of traffic, and higher cost relative to LAN links; BTOP has the potential to eliminate that distinction.

18

The reporter in question has plenty of clue.

19

Hello every one

I am curious as to how others are documenting their network; both visually and configurations.

Is there any a software offers a database with web-based front end that can document in a very details.

thanks

20

One successful BTOP application in North Carolina has more to do with enterprises (in this case, educational institutions) than with consumers.