Thursday, March 26, 2009

SIP to Share


I originally wrote this entry on September 17, 2004, and published it on blogs.sun.com.



A minimal knowledge of the Session Initiation Protocol (SIP) should make it obvious how it can be used to build a file-sharing system.


By the way, SIP is finding its way to Solaris, too.

The Network is the Computer

I originally wrote this entry on September 15, 2004, and published it on blogs.sun.com.


In my mind, there's no more revolutionary concept in computing, networking and information technology than the motto which Sun coined in many of its corporate PR campaigns: The Network is the Computer. The origin of the motto, within Sun, remains unknown to me, but I would sure like to discover it by some piece of corporate archaeology. (I'm sure we have our un-official, as well as official, archaeologists here who know the answer.)


I can even imagine a new PR campaign based on the motto--a TV advertisement perhaps: A large number of sleepy and tired workers in cubicles are running routine errands of the most stifling kind; the beautiful jumble of the New York skyline can be seen in close view and is visible through the wall-length windows but no one is paying any attention to it; a rumor begins to spread from a remote corner of this vast room; "The Network is the Computer," whispers someone as if awakened with new life; as the "rumor" spreads throughout the room (the building and the town, in the later frames), the mood swings to jubilation and true excitement--the revolution is here. The last frames focus on a person who, the audience can guess, may have something to do with the rumor--a young engineer with a Sun T-shirt on. [That would be a cool ad ! Perhaps, I should receive some sort of compensation for designing it! (Please excuse my indulgence. My only sin is that my father was an advertising executive in Iran in the mid 1970s, and he did take me to work a few times.)]


Many others, including Tim O'Reilly, have opined on the motto.


To me, it has an almost esoteric meaning, and I'm fond of such esoterism:


  • The only computer that matters is the network.
  • The network is equivalent to one giant computer with multiple entry points. Ultimately, it is equivalent to a single Turing machine. (Or is it? What about external, interacting "machines". Surely, their purpose could not be modeled as merely random.)
  • The only computing that matters is the one that make the network more effective and efficient.
  • Those that claim the desktop to be the (or a?) computer have gotten it totally wrong.



To you, I'm sure the motto could mean something quite different, but if it could mean different things to different people within Sun, how could it be a component of its corporate identity or its organizational purpose? The answer is probably that, in fact, there's a great deal of commonality in how people at Sun understand the motto: The Network is the Computer.

 

Spam As Failure Affords No Finite Solution

I originally wrote this entry on September 9, 2004, and published it on blogs.sun.com.


It looks to me that there's actually no absolute or fail-safe method of preventing spam.


This sort of claim is bad marketing if one wants to sell a product that claims to do just that, but I'm sure product limitation discussion will surface if the buyer is sophisticated enough to ask some of the right questions.


As long as the network is useful, there will be those who will use it "improperly." Let's define spam as the improper uses of the network to generate unwanted content. (There are other, more limited definitions of spam.)


Furthermore, if we think of improper uses of the network as failure events, we may be able to apply some work in termporal logic cited in Ken Birman's book (Building Secure and Reliable Network Applications), which basically says that as long as we can always add one more failure recovery method, we will be able to recover from all failures.


In other words, there is no finite solution to the failure problem, in this case spamming as defined above. Of course, this is no excuse to deploy operating systems and computing environments that are more prone to such attacks. Let's face it. Even if the problem has no finite solution, there are environments that are more secure than others. A system's level of security and protection is based on the effort required to break it.



My claim regardging the infinitude of the spam problem is similar to saying that biological viruses are always going to be around.


Viruses are made to take "improper" advantage of the extreme usefulness of certain biochemical structures including self-replicating macromolecules. To fight viral and bacterial attacks, our bodies use the versatile, highly-tailored antibodies to defend our physical perimeter.


I think those who wish to make "proper" use of the Internet should probably use similar methods to fight spam and other improper uses of the network.


The Global Rise of the Chinese Network Equipment Providers

I originally wrote this entry on September 9, 2004, and published it on blogs.sun.com.


If you have an online or hard-copy subscription to the Wall Street Journal and deal with the global telecommunications equipment market, you may want to take a look at the article by Christopher Rhoads and Charles Hutzler on the global rise of the Chinese telecommunications equipment providers, including Huawei, Fiberhome Communication Technology Ltd. and Zhongxing Telecom Ltd. The article appeared this Wednesday, Sept. 8, 2004, on Page One of the WSJ.

Information or Telecommunication Service?

I originally wrote this entry on August 30, 2004, and published it on blogs.sun.com.


The government has appealed to the Supreme Court a Ninth Circuit Court ruling on cable services that required the FCC to regulate them as telecommunications services.



Last October, the Ninth Circuit, based in San Francisco, found unwarranted the FCC characterization of cable high-speed Internet services as unregulated "information service."


The court ruled that cable-modem services should be regulated more like a telecommunications service, which is more heavily regulated by the Federal Communications Commission. (See, for example, the unbundled network element requirements imposed on telecommunications service providers. The requirements have briefly been discussed on this very weblog: 1 and 2.)


The Ninth Circuit Court's decision came from a lawsuit filed by a small Internet-service provider called Brand X, which sued after the FCC developed rules that only lightly regulated cable broadband, (FCC v. Brand X).



"The high court will add the case to its docket and will likely make a decision on the appeal sometime late this year," reports Mark Anderson of the Wall Street Journal.




The Ninth Circuit is very active in rulings on communications services. (See the material on music sharing on this very weblog: 1 and 2.)

To Share or Not to Share, That's the Question

I originally wrote this entry on August 26, 2004, and published it on blogs.sun.com.



On August 19, 2004, the 9th Circuit Court of Appeals had ruled in the Grokster case that the makers of peer-to-peer software were not responsible for what users do with their network. The list of plaintiffs vs. defendants is quite interesting to review. Here are some significant paragraphs from the BBC report on the case:


In their ruling, the judges said the case had parallels with older cases which said video recorders should not be banned just because some people put them to illegal ends.


"History has shown that time and market forces often provide equilibrium in balancing interests, whether the new technology be a player piano, a copier, a tape recorder, a video recorder, a personal computer, a karaoke machine, or an MP3 player," wrote the judges in their opinion.


The judges said it should be up to Congress rather than the courts to change copyright laws.


One factor that led the US court to rule in favour of Grokster and StreamCast was the lack of central servers that co-ordinate file-swapping activity.


In the past Napster's use of central servers led the same court to call for that network to be shut down.


"Today's decision should not be viewed as a green light for companies or individuals seeking to build businesses that prey on copyright holders' intellectual property," said Jack Valenti, MPAA chief executive.



This is a reasonable analysis.


Only a few days later, on August 26, 2004, reports said that U.S. FBI agents had raided five homes across America as part of the government's first federal crack-down on file sharing networks. The five peer-to-peer file sharing hobs raided in Wisconsin, New York and Texas operated on Direct Connect software. DoJ siezed equipment but made no arrests, reported the BBC.



The DoJ raid seems to be somewhat at odds with the Grokster ruling. In other words, DoJ is breaking up an "innocent" private network rather than arrest or punish anyone misusing it. Perhaps, an arrest will lead to legal arguments where the concept of fair use will have to be finally settled in more detailed outlines.


Richard Posner has written briefly on the interaction of technology and law in the Grokster case as well as on fair use.


Spam, 3/4 of e-mails in the U.S., costs $10billion a year

I originally wrote this entry on August 25, 2004, and published it on blogs.sun.com.


The Associated Press reports, form industry sources, that "spam e-mail accounts for almost three-quarters of the e-mail in the U.S. and costs consumers and businesses as much as $10 billion a year."


Pricing Unbundled Network Elements

I originally wrote this entry on August 20, 2004, and published it on blogs.sun.com.


I wrote earlier about the Telecom Act of 1996, focusing on its provisions regarding unbundled network elements.


Pricing of such elements has been a point of contention between the FCC, the local exchange carriers and their competitors. The FCC has required the incumbent carriers to provide their elements at a discount, while the carriers prefer a pricing model that emphasizes current costs of maintaining the network.


This past Friday (August 20, 2004), the FCC forze prices incumbents charge to competitors for six months.


Some say this has been done at the request of the administration, to prevent price hikes prior to the upcoming presidential elections. The 3-2 vote was on party lines. Anne Marie Squeo of the WSJ reports that


Bell officials have repeatedly said their interpretation of the March appeals-court ruling, which went in their favor, is that they no longer have to lease all the various parts of their networks at discounted prices to rivals. The companies contend that the wholesale rates don't cover their costs for maintaining the networks. Rivals, who repackage the Bells' service and sell it to residential and business customers, have taken more than 19.5 million customers from the Bells through such arrangements during recent years.



The differences over the 1996 telecom act are sure to conintue. If the network operators are not allowed to charge real prices, they will not have the incentive to upgrade their networks with new equipment and will have to resort to increasing prices on their own direct customers. Unreal low pricing for one segment will lead to unreal high pricing for another segment of the market. This is a situation FCC has to deal with. However, institutional politics behind its decisions makes matters no easier.



Sunday, March 22, 2009

Pricing Unbundled Network Elements


I originally wrote this entry on August 23, 2004 and published it on blogs.sun.com.


I wrote earlier about the Telecom Act of 1996, focusing on its provisions regarding unbundled network elements.


Pricing of such elements has been a point of contention between the FCC, the local exchange carriers and their competitors. The FCC has required the incumbent carriers to provide their elements at a discount, while the carriers prefer a pricing model that emphasizes current costs of maintaining the network.


This past Friday (August 20, 2004), the FCC forze prices incumbents charge to competitors for six months.


Some say this has been done at the request of the administration, to prevent price hikes prior to the upcoming presidential elections. The 3-2 vote was on party lines. Anne Marie Squeo of the WSJ reports that


Bell officials have repeatedly said their interpretation of the March appeals-court ruling, which went in their favor, is that they no longer have to lease all the various parts of their networks at discounted prices to rivals. The companies contend that the wholesale rates don't cover their costs for maintaining the networks. Rivals, who repackage the Bells' service and sell it to residential and business customers, have taken more than 19.5 million customers from the Bells through such arrangements during recent years.



The differences over the 1996 telecom act are sure to conintue. If the network operators are not allowed to charge real prices, they will not have the incentive to upgrade their networks with new equipment and will have to resort to increasing prices on their own direct customers. Unreal low pricing for one segment will lead to unreal high pricing for another segment of the market. This is a situation FCC has to deal with. However, institutional politics behind its decisions makes matters no easier.



Unbundled Network Elements


I originally wrote this entry on August 16, 2004 and published it on blogs.sun.com.


Transaction costs include legal costs of maintaining a regulatory regime. A good regulatory regime should optimally reduce such costs.


On Lessig Blog, Tim Wu writes that "the Telecom Act of 2006" should include among other things the abandonment of 1996 Act's "Unbundled Network Element" approach to telephony competition. "The litigation costs just aren't worth it," he adds.


With rather great detail on particular cases of unbundling, the Telecommunications Act of 1996 stipulates that operators would have to make available, for sale to their competitors, their unbundled network elements. This implies unbundling not only physical elements but the uses of that physical element. So, for example, operators owning voice and DSL service infrastructure, such as the last mile of connectivity, would have to make the infrastructure elements for these two services available to their competitors as an unbundled, separate and individual market items for each service type.


To be more precise, the 1996 FCC Act, defines and stipulates unbundled access up to the point such that the elements could be rebundled by the buyer.





UNBUNDLED ACCESS- The duty to provide, to any requesting telecommunications carrier for the provision of a telecommunications service, nondiscriminatory access to network elements on an unbundled basis at any technically feasible point on rates, terms, and conditions that are just,
reasonable, and nondiscriminatory in accordance with the terms
and conditions of the agreement and the requirements of this
section and section 252. An incumbent local exchange carrier
shall provide such unbundled network elements in a manner that
allows requesting carriers to combine such elements in order to
provide such telecommunications service.







VoIP, A New Legal Regime


I originally wrote this entry on August 12, 2004 and published it on blogs.sun.com.








VoIP


regulations are likely to be revamped, writes Rep. Rich Boucher of Southwest Virginia.


Rep Boucher, guest blogging at LessigBlog, describes some of his efforts in the Congress to revamp VoIP regulations within the context of a new regulatory regime to follow the Telecommunications Act of 1996.


Under the ideal circumstances, all such regulations should primarily develop possibilities for commerce in ways that best reduce transaction costs.


Wireless Grids


I originally wrote this entry on August 9, 2004 and published it on blogs.sun.com.


The July / August 2004 issue of the Internet Computing focuses on wireless grids. (You need to have a subscription to view all the articles but the guest editors' introduction is available online.)


Just like other grids, wireless grids are a type of resource-sharing network. As the introduction points out


Following Metcalfe's law, grid-based resources become more valuable as the number of devices and users increases . . . In some ways, wireless grids resemble networks already found in connection with agricultural, military, transportation, air-quality, environmental, health, emergency, and security systems. A range of institutions, from the largest governments to very small enterprises, will own and at least partially control wireless grids. To make things still more complex for researchers and business strategists, users and producers could sometimes be one and the same. Devices on the wireless grid will be not only mobile but nomadic--shifting across institutional boundaries.



The guest editors identify three classes of applications:


  • Class 1: Applications aggregating information from the range of input/output interfaces found in nomadic devices.
  • Class 2: Applications leveraging the locations and contexts in which the devices exist.
  • Class 3: Applications leveraging the mesh network capabilities of groups of nomadic devices.



The authors also note three salient features of wireless grids that help distinguish them from wireline and traditional, computer grids.




  • new (grid) resources,
  • new (non-traditional) places of use, and
  • new institutional ownership and control patterns.



Most of these features exist because the vast majority of wireless devices are currently mobile devices. These devices may be geo-location-aware and can cross various geographic and institutional boundaries.


The guest editors' introduction is quite a good read. It outlines the types of resources wireless devices bring to networks and tracks the institutional changes, from centralization to decentralization, which have led to the evolution from the mainframe to the PC and to the handheld.


Note that this evolution, particularly the move from PCs to handhelds does not mean that servers are a by-gone conclusion. Much of the smarts in handhelds can only be deployed in real world scenarios if they are connected to a wealth of server-based content, applications and intelligent networks. So, servers are here to stay even if the "decentralization" process the authors hold as responsible for the evolution to mobile devices continue to persist.




Network Identity and Access Management


I originally wrote this entry on August 6, 2004 and published it on blogs.sun.com.


When Network Computing magazine starts to roll out the carpet for network identity and access management, one should conclude that the market is finally noticing the importance of network identity and access technologies.


Nevertheless, we can still wonder why it took so long for the world to realize the significance of what Sun had been saying for years. (After all, the Liberty Project has been around for a long time.)


Network identity and access management is a fundamental technology for mobile and enterprise environments where sophisticated, aggregated value-add services are integrated from a set of distinct business and functional entities.


The initial market delay could probably be attributed, at least in some part, to the initial wavering by some big ISV's which later melted away as large customers demand for the technology became more and more obvious. As the mobile world became set to go with network identity and access management technologies, some of the traditional technology providers who were sitting on the sidelines became much more active. All of the major IT companies are now active in the Liberty Project.


Note: Thanks for the comments on this post. Originally, I had stated some of the material in the last paragraph, most incorrectly, in the present tense. I've tried to make some necessary corrections on the post to fix these.


Wiretapping VoIP and PoC


I originally wrote this entry on August 5, 2004 and published it on blogs.sun.com.


In a 5-0 preliminary ruling, the FCC has extended the wiretapping requirements of CALEA (Communications Assistance for Law Enforcement Act) to VoIP (Voice over IP) and PoC (Push-to-talk on Cellular).


(The preliminary rulling does not affect, non-service-provided, peer-to-peer VoIP.)


See Tim Wu's blog for a brief on the economic and legal issues the ruling raises.


See Declan McCullagh and Ben Charny's report on news.com for a report and related material.


ACLU, Americans for Tax Reform, and Center for Democracy and Technology have already expressed their reservations against extending the CALEA to the Internet service providers.


What I want to highlight here briefly is some of the technical difficulties of wiretapping on the Internet as opposed to the traditional PSTN networks.


The major difficulty arises because of the built-in routing flexibilities in the Internet. In other words, what makes the Internet resilient to local failures also makes it harder to wiretap.


Packets can take different routes. For example, RTP (the real-time transport protocol), the most common protocol for conveying VoIP packets, does not require a reservation model along the lines of RSVP. The main thing about RTP as compared to TCP is that it does not retransmit "droped" or "lost" packets composing its "voice" or other media payloads.


Some of these difficulties of wiretapping could lead to business models where consumer devices have modules capable of actively participating in wiretapping. These business models are broken from the start.


Requiring that such devices be put in the consumer's hands may discourage use of the Internet and make it more expensive. Furthermore, alternate non-conforming but smart devices could be installed to defeat the purpose of wiretapping or to skirt it all together.


Taking the non-device-dependent approach, the level of coordination required for wiretapping on the network, through firewalls and other intermediaries, not to mention through a variety of routers and switches, is truly mind boggling.


Wiretapping solutions (independent of end-point device participation) will only be available at considerable cost. (As recorded in Declan McCullagh and Ben Charny's report on news.com, Verizon Wireless' lukewarm response to the FCC's preliminary ruling confirms these difficulties.)


Same technical issues and difficulties hold for Wiretapping PoC.


Finally, while congress appropriated $500 million to reimburse traditional PSTN phone companies for CALEA compliance, such compensation will apparently not be available to VoIP and PoC providers.