|
North American Network Operators Group Date Prev | Date Next | Date Index | Thread Index | Author Index | Historical Re: floods
Dear Hoisl, ] I am doing a project for school and I was wondering if you could answer my ] question. What can be done to reduce the amount of flood damage? Thanks. As someone that has years several years of my life working on the Internet, I'm proud to see it being put to such use. As well, I'm proud that our American schools are teaching routing protocols. Thanks for asking such an intriguing question. Contrary to popular belief, flooding is not necessarily a bad thing. Certainly, unicast routing updates contain less gross overhead for the network in question. Nonetheless, the simplicity of RIP is a win for many network operators in today's internet. As well, many advanced IGP's in use on many of today's largest Internet backbone providers use OSPF and ISIS as an IGP, both of which utilize flooding. From another paradigm, email floods sent to people who do not want it is often considered flooding, though 'Spamming' is a more popular term. I will address this issue towards the end. With regards to RIP, and to absorb a subset of this question, we certainly do absorb damage from this flooding mechanism. This is not inherent to the protocol, but by the use of said protocol in distributing igp routes into the egp, BGP4. Certainly the periodic properties are most alarming, especially with their systemic intrusions into the backbone BGP table. Various theories have been formed regarding this 30 second periodicity, including one lovely old mentor's of csu/dsu timing. At first glance, one may wonder why an advanced routing protocol such as ISIS or OSPF floods their packets all about the network. I wonder this myself, though I am chagrined to note that I didn't ponder these thoughts until well after high school. I believe that the simplicity inherent in the transport of routing updates is a significant benefit. Just as a simple pair of scissors makes a wonderful tool due to its few moving parts, flooding removes a dependency upon the routing innards to absorb, process, and selectively forward the updates. As well, while adjacency formation is dependent upon proper SPF calculation, adjacencies are of little benefit in propogating LSPs, other than to provide paths. Certainly ISIS and others peer inside briefly to prevent loops, but as the SR bit is all that's checked, I'm certain you can observe the simplicity. Mutation of the update is reserved for more complex protocols. With regards to electronic mail, this has become a large issue. Hardly a day goes by that I receive no unwanted email. And let me tell you, some of the content makes me blush and giggle! Nonetheless, the slight overhead on our Internet, as well as the hundreds of man hours required across the world to press <D>elete significantly detract from the goal of our work -- decrease communication barriers and increase productivity -- are serious problems. Various Laws of Thermodynamics and Chaos study come to mind here, but I'll leave it to say it's not unexpected. The world evolves into all mediums, and this is entirely predictable. What we need is your help -- write letters, stop the flooding! Send them to everyone! Nonetheless, we've several SandBaggers that are to be lauded. Their work in combating these evil greedy ... mean people is helping to raise awareness. And this is important. In conclusion, flooding does damage our network. Ideally our world will evolve into dynamic SVCish contacts with routing/switching entities negotiating hierarchy and position autonomously. With regards to email flooding, the overhead in dealing with this is high, and certainly causes damage. Good luck to you in your studies, and as your routing protocol understanding increases, please feel free to look me up. -alan - - - - - - - - - - - - - - - - -
|