William H. Magill on Mon, 22 Sep 2003 12:38:09 -0400

[Date Prev] [Date Next] [Thread Prev] [Thread Next] [Date Index] [Thread Index]

Re: [PLUG] Good, inexpensive co-location

On Sunday, September 21, 2003, at 08:37 AM, W. Chris Shank wrote:

Great, let's put it all in the same place. Smart thinking.

It is the nature of the communications infrastructure in the United States, and for that matter, world wide. It doesn't matter what scheme you use -- Cable, Wireless, Land line, it's all built on a star topology -- everything aggregates to a central location and those central locations may or may not connect to multiple "up-stream" locations.

This is why TCP/IP was developed as a "CONNECTIONLESS protocol" -- the assumption was, knowing that there were many "NAPs" (as we call them today), the packets could travel via many different paths, if those paths existed, independent of physical connectivity.

But remember, something like 98% of so-called "redundant" connections are not -- that is why there was such a massive problem a couple of years back when an Amtrack backhoe tore up the fibre in Princeton Junction. ... it seems that all of the Backbone carriers used the railroad right of way as the best place for their Fibre. You may buy your connection from a different ISP, but they just use MaBell for the "last mile" -- so your redundant connections sit right atop the same pole, and all come into your building through the same hole in wall.

One assumes that you know the derivation of SPRINT -- The Southern Pacific Communications Company. The name (and principals) has wandered around from Brown to United to US Sprint to Qwest over the years... but SPRINT laid the first Fibre along its RR right of way back in the 80s and then leased that dark fibre to everybody from MCI to GTE and so started the major move to lay Fibre along Rail Road rights of way. Why? Because the RR rights of way when right into the heart of every major city in the country and therefore the hassles were nonexistent.

Similarly, today, the LARGEST fibre plant in the Delaware Valley, belongs not to any telecommunications company, but the the Electric Company -- PECO Hyperion. They originally built the fibre plant to interconnect all of their substations in the region, using THEIR rights of way. They don't provide "last mile" connectivity in the Delaware Valley, but they do in, I think it is, Virginia (or may be Maryland).

And 401... it was built as the Packard Manufacturing company atop 4 sets of rail road tracks. They built the cars and trucks in 401 and shipped them out via rail or sold them from the sales room next door.

Locations like 401 know that they are aggregations and therefore can spend the money needed to provide REAL UPS support. The "battery backups" which people send money to APC for are good for one thing only -- they allow you to shutdown when the power fails. They do not allow you to keep running if your primary power feed fails for more than 10 or 15 minutes. But generator driven UPS units cost big bucks to install and maintain, normally as much as, if not more than, the Main Frames they support. And facilities smaller than main-frame operations simply cannot afford them. 401 might have come into existence "location-wise" surreptitiously, but not without much planning and thought.

William H. Magill
# Beige G3 - Rev A motherboard - 768 Meg
# Flat-panel iMac (2.1) 800MHz - Super Drive - 768 Meg
# PWS433a [Alpha 21164 Rev 7.2 (EV56)- 64 Meg]- Tru64 5.1a

Philadelphia Linux Users Group        --       http://www.phillylinux.org
Announcements - http://lists.netisland.net/mailman/listinfo/plug-announce
General Discussion  --   http://lists.netisland.net/mailman/listinfo/plug