Josh King's Field Reports

PSO Post-Mortem

Josh King — Wed, 16 Jul 2008 21:42:22 -0600

So, first of all, it was great hanging out with everyone at the PSO in Boston this year. We have a really great crew this year, which makes it a fun and challenging time to become a VISTA Leader, and I'm really looking forward to the next 12 months.

Since the PSO, I've been mainly going over the PSO feedback, getting caught up on my projects at Acorn Active Media, and helping to get Stephen and Nicole settled in their roles as VISTAs at the Urbana-Champaign Independent Media Center, which is about 4 blocks from my apartment. Acorn is closely allied with the IMC, and I've always spent a great deal of time there helping out. It's really great to see the IMC having regular staff meetings, and mobilizing around getting them active on their projects.

I have to go up to Chicago this weekend, so right now I'm just trying to make sure that the two of them have everything they need, and to complete some of my projects before I have to leave. Currently I'm working on creating new drafts of the research whitepaper on open-source wireless mesh networking that I completed in May, and of the ARIN IPv6 policy proposal for community networks. The former needs to incorporate some edits by another member of Acorn, and I'm using the opportunity to shore up some of the citations and to add screenshots from the newest build of the mesh software we're developing, but in order to do that I need to build custom images and determine how to flash them onto the hardware I have (Fonera+'s). The latter was introduced at the ARIN meeting in May, where it received a lot of interest but was sent back for revisions. There's a advisory council meeting on a Thursday in the middle of each month, so I was trying to get the edit in for this Thursday, but I'm not sure if I'll make it with all of my other projects. It has to be done by September at the latest.

I'm also working on a website design project for Tokyoprogressive.org and ongoing server maintenance on the Chambana.net colocation project. Chambana.net has a new mailserver that I want to get online, and set up filesharing, distributed logons and a local Ubuntu mirror for the production lab that my VISTAs are working in. Hopefully I'll be able to get at least some of these done before Friday, but if not I should get them done in short order next week.

IPv6

Josh King — Thu, 29 Nov 2007 16:58:34 -0700

So, long time no post. I was going to talk about one of the things I've been working on, which is IPv6. This'll take a little bit of background explanation, so if you're already tech savvy and know all about IPv6 just bear with me.

You probably have at least some idea of how the Internet works. Virtually every computer or device connected to the Internet (or just about any computer network really) is assigned an address, known as an "Internet Protocol" (or IP) address. These IP addresses are usually represented by a string of four numbers between 0 and 255 separated by dots, like 74.134.239.209. Computers on a network send packets of information much like you or I would address a letter. The packet is addressed with the number of the computer that it wants to send the information to (like a web browser requesting a copy of a webpage) and sends the packet on its way. That packet makes its way to the target computer, which reads the information and puts it to whatever use was intended (like processing the correct webpage and sending it back to the web browser). Exciting, eh?

The version of the Internet Protocol that we use is known as version 4, and has been in use since 1981, and has worked well. However, we are in perilous danger of exhausting the total number of unique address numbers available. There are only 4,294,967,296 unique addresses available using the current standard. This might seem like a lot, but consider that it is not even close to as many addresses are there are people in the world (~6.5 billion). Now think of how many Internet-connected computers and devices there are in the world, each of which possess one or more IP addresses. Furthermore, there are millions of those addresses which are reserved for special purposes, like private networks, multicasting, and other network-architecture related functions. So our available address space suddenly seems quite small. In fact, current estimates of address exhaustion place the time that we run our of unique addresses at sometime in the first half of the year 2010, at which point without a solution architectural growth of the Internet will come to a halt. This situation is what is known in technical terms as Not Good.

Enter IPv6. IPv6 is the next version of the Internet Protocol (IPv5 was an experimental streaming protocol that was designed to complement IPv4 but was never used). In the words of Bill Nye, IPv6 is Really Bigly Hugely Big. The address space is so large (2^128 addresses) that it's difficult to even adequately describe, so I'll attempt to do so with a series of metaphors I found on Wikipedia and elsewhere. IPv6 has roughly 50 octillion addresses (a 5 with 28 zeros after it) for each person on Earth. If the population of the Earth were 18 quintillion, there would be about 18 quintillion addresses for each of them. If the IPv4 address space were the volume of an iPod, then the IPv6 space would be the size of the entire Earth. I especially like this one: there are a million times as many IPv6 addresses as there are IPv4 addresses...for each star in the known universe.

So you get the idea. IPv6 is big, and should last us for a while. It also solves a lot of problems that IPv4 has introduced. In order to stave-off address exhaustion, various network technologies have been introduced which break the end-to-end nature of the Internet. Originally, the intent was for the Internet to be the great non-hierarchical redundant network, with every single Internet-connected computer (or "Host") with an unbroken two-way path to every other host. However, technologies like Network Address Translation (NAT) have been introduced which allow hosts to share "public" (visible to the rest of the Internet) addresses. This is why if you've ever looked at the network settings on your laptop when you're out at a café or something, your address always seems to be 192.168.blah.blah or something similar (maybe 172.16.blah.blah). These are addresses set aside for "private networks," which are networks which are never visible to the rest of the Internet. Probably the café only has one IP, which it shares with all of its clients. The end-to-end path is broken, because a set of hosts are behind a hierarchical structure made by NAT. Often there is a similar situation at home, where all clients of an ISP are behind the ISP's NAT. This makes it difficult if not impossible for you to publish content to the Internet without utilizing third-party services and websites, or paying ISPs extra money for "business class" services on top of what you pay for bandwidth.

What's more, that one address is probably "dynamic" rather than "static." That means that the café's Internet Service Provider (and probably your ISP at home) assigns an address to them dynamically from a common pool as they become available. Why is this a problem? Well, it's not, unless you want to provide content, like a website. A protocol called Domain Name Service (or DNS) assigns human readable addresses (like ctcvista.org) to IP addresses (like 208.113.151.31) based on a database of those pairings. If ctcvista.org had a dynamic address, then if the ISP decided to change it to 208.113.151.32, ctcvista.org would no longer point to the right place and no one would be able to find the website. This means that in order to have a website, people have to pay hosting providers for space on computers that have expensive static IPs that don't change.

Still with me? Okay, so IPv6 can solve these problems by assigning a public, static IP to every device on Earth. However, who gets to assign the addresses? Currently, all the IP addresses on Earth are assigned by the Internet Assigned Numbers Authority (IANA). IANA delegates that responsibility to organizations in charge of large geographical areas. So all of the addresses in North America are assigned by the American Registry of Internet Names and Numbers (ARIN). ARIN allocates ranges of IP address numbers called "blocks" to qualifying organizations, mainly large ISPs. This makes sense in a way, because ISPs provide the connectivity and it is architecturally simplest for the assigned numbers to reflect the actual structure of the networks that make up the Internet.

However, there are a large number of community projects that require a large number of addresses but don't fit into the established model. These are projects like community networks, municipal wireless networks, community technology centers, and community-run ISPs. These groups are often short on cash but need a large number of addresses for all of their computers, devices, and members. But even though there are a huge number of addresses available, they still have to pay an arm and a leg to go through the large ISPs that maintain a stranglehold on the address space as if it were still running out.

So, finally, enter my organization, Acorn. Acorn is working with ARIN to do two things:
1. Get a large allocation of addresses that Acorn can then hand out to community projects that need them without them having to go through the big ISPs.
2. Change the policy so that in the future any other organization can do the same thing.

Number one is accomplished. Acorn's been awarded a /32 block, which is 2^92 addresses. Billions and billions, to say the least. Now we're working on a policy proposal for early next year which, if ratified, will set up a procedure for other community organizations to get their own IP addresses to provide services to the people in their neighborhoods without having to go through the ISPs for anything but connectivity. Hopefully with this work, we can make the Internet become a better, fairer place like it was intended to be before the critical date in 2010 before IPv4 runs out.

So I hope that this is a fairly straightforward explanation while still being accurate. I'm far from being the world's foremost expert on Internet architecture, so if I got any points wrong feel free to correct me.

technobabble & foodstamps

Josh King — Mon, 20 Aug 2007 11:05:01 -0600

It's been a long time since I wrote a field report (I'm really going to try to be more attentive to this). There are a lot of projects that I'm currently working on. I'm trying to do a a complete redesign of my organization's website, in order to integrate a lot of project management and accounting stuff into a centralized location, since my organization's staff is a little far-flung. The current site is running Drupal 4.6.3. I spent a week trying to manually upgrade it in stages to Drupal 5.2, but ran into so many database errors that I eventually created a new test site in 5.2, with the intention of getting that stable and then porting the content. I highly recommend the module casetracker (http://www.drupal.org/projects/casetracker). Most ticketing systems are designed for software development, but casetracker is much more generalized, so you can have general tasks that have support tickets submitted against them. I'm trying to integrate that with Organic Groups and a wiki (wikitools+pearwiki+views), so that projects created in casetracker can have a group assigned to that project, and each task within the project can be associated with one or more pages in a documentation wiki. I'm also trying to integrate the ERP module, which does complete business management, so that tasks can be assigned hours and values that are automatically entered into the timesheet in the accounting software. If I can get this all to work I'll post a complete tutorial on the CTC VISTA wiki, if people are interested.

Besides that, I'm still trying to get the cool but somewhat buggy distributed monitoring software ZABBIX (http://www.zabbix.com) working with our datacenter. Partly I think it's our firewall setup, but I think it'll help once we get our new colocated server in place (the current virtual server where I'm setting it up is somewhat overloaded). Once that's setup, our servers will be able to alert us when there's a problem, rather than us having to go look for it.

I'm going to be writing a whitepaper soon on the ideal open-source wireless mesh solution, and I'll be doing some hands-on research with some state-of-the-art equipment to do it. I'm also going to be working on an application to ARIN for IPv6 address space allocated specifically for community networks (never too early).

There are three other especially big projects that I'm working on, one of which is probably going to take up the majority of my time during the next month. I'm not sure how much I'm supposed to say about these projects yet, unfortunately, because of the groups involved, but they'll be exciting to talk about once they're a little further along.

Also, I finally managed to get foodstamps. Apparently in Illinois, Americorps income is exempt from considerations as far as how much you get for foodstamps, so I got the maximum amount ($155/month). This will be a huge help, because besides rent, food is my number one expense. Things are busy but going pretty well, I'll try and post these much more regularly from here on out.

Working hard on things that are hardly working

Josh King — Fri, 13 Jul 2007 13:24:08 -0600

It's been an uphill battle this week with our internet service provider, who has been doing maintenance in this area resulting in major outages for Acorn's Chambana.net datacenter, along with all of the other network services in the building. Between me and my supervisor we've probably spent about 4-5 hours on hold with tech support over the last couple days. There's been a lot of talk about switching to some other ISP instead of Insight, but most of our options are significantly more expensive for little more capacity than what we already have. We are still hoping to get a deal with the city of Urbana for a fiber connection into the building, but that seems to have stalled for the time being. On the bright side, we've managed to find a couple of really nice 42u racks for only $100 a piece (these are worth more like $800) that belong to a friend of the project, and so we're going to be moving those into our server room soon, and it'll be a lot better than having our servers sitting on crappy wire shelves or on the floor. Still a lot of work to do getting Chambana.net up to spec, but I hope by the end of the year we'll have completely fast and redundant hosting services on real racks with gigabit ethernet, a fiber connection, battery backups, good electrical, physical security, HVAC, and good distributed management/monitoring capability. We hope to have our main website updated in the next couple of weeks, and I'm going to try and integrate all of our services into that drupal installation, including a donations page and a ticketing system.

On the wireless side of things, soon we're hopefully going to be getting a research grant from the NSF. We'll be able to do a lot of development and even a little hardware deployment with the money we'll get. We've also got a tentative deployment plan with the city of Urbana that will deploy three more nodes, after which we'll have all of the outdoor seating areas in the downtown provided with public wifi hotspots. We need to move fast on this because Urbana's adjoining city of Champaign just announced a deal with a company to deploy a proprietary wireless tech in their downtown, and we don't want them to beat us to the punch when we've been working on our grassroots OSS solution for the last three years.

Finally, got my loan forbearance and am applying for foodstamps, so I should hopefully squeak by at the end of the month here.