Eskimo North

Eskimo North Hardware - July 11, 2001






Communications Servers

Because 56k dial-up requires an entirely different infrastructure and to obtain an international reach, we have chosen to use various port wholesales to provide 56k and ISDN dial-up service rather than inhouse hardware.



Network Equipment




Future Plans

I write this with a lot of trepidation because it seems the older I get the more cloudy my crystal ball becomes. The last time I updated this section was a little more than two years ago and although we've accomplished many of the goals of intended upgrades, the way it was done turned out to be quite different than I had anticipated.

In 1998, we had anticipated adding SSL and database capabilities to the web services here, those things have been done. We now have a database and an SSL server, although I intend to replace the proprietary Raven server with Apache with the SSL mod.

We anticipated needing to increase the size of the news spool and did so by roughly a factor of ten, still the growth in News volume has totally outstripped that. We're now in the process of putting together an entirely new server to address capacity issues with another ten-fold increase in spool capacity.

We entered the DSL market using a reseller (MegaPOP/StarNet Inc) that abrubtly terminated services and left our customers and us stranded. We were unable to find a reseller to replace MegaPOP (and it's not that I didn't look). We are now in the process of establishing a connection to Qwests DSL cloud and will be providing our own data backhaul rather than depending upon a reseller. This should be a killer arrangment though when it's up and going as there will be only one high performance router between the DSL cloud and the Internet backbone making for ultra-low latency and high security since there will be no point where the data crosses a LAN and can be sniffed.

In 1998 we had just gone through a phase of taking a handful of machines and replacing them with a much larger number. We are now going the opposite direction, consolidating functions on fewer machines. This has become desirable because the CPU, memory, bus, and disk I/O capabilities of modern machines have made it possible and fewer machines are easier to administer and create fewer opportunities for communications problems between machines which in the past have been the basis for many reliability problems.

In 1998 I had anticipated the need for more router capacity and had planned to upgrade the routers processor. We have done so, going from a single 20Mhz CPU to a dual 80Mhz CPU and now have very adequate capacity. The router can handle four T1's and a FDDI running heavy traffic at about 20% CPU occupancy.

We now have three T1's through Sprint, using two different POPs and have redundant routing so the failure of a circuit or even an entire Sprint POP does not isolate us from the Internet. We use multiple port wholesalers for dial access which provides redundancy in most major metropolitan areas.

In 1998 I had anticipated a move toward IPv6, and we upgraded router software and have been migrating to Linux in preparation for that. But thus far Sprint is not routing IPv6 and so even though we have the local infrastructure largely in place (the SunOS boxes will not be capable of IPv6) we can't utilize it.

I had anticipated a growth in broadband access, and it has happened. The bulk of the broadband market has gone to cable modems and this has presented some unanticipated challanges. Cable modem providers often provide only one IP address per customer. This has forced the use of NAT which breaks many protocols such as identd and this has forced us to re-think the way many of our services are provided in order to work around problems that NAT introduces.

The world-wide web has emerged as a globally transformative phenomena and I continue to be concerned that you get to stay in the loop rather than just becoming a "consumer" of mass-media. The very largest challange I see in this area is making it easy enough for someone who does not have a PHD in computer science to be able to publish on the web.