Tag Archives: uptime

Building a crash-proof internet, Off-the-Internet Processes

Interesting article in NewScientist speaking towards the challenges of building a crash proof internet.  Bennett Daviss provides accurate information regarding the challenges of the internet, and how it has become a mission critical part of our lives – personal and professional.  The Internet is not guaranteed to be up and unless conscious effort is taken to ensure that your business’ packets are flowing it is likely a random event will cause a disruption of at least one hour if not many hours.  RackSpace’s operational challenges the other day highlighted this fact.

The article has a nice breakdown on the threats and highlights a specific solution being revamping the routers.  In order to achieve this ‘revamp’ it is necessary to deploy new and emerging concepts onto in-production devices without causing an interruption has led to the need for a separate test bed.   The concept of building a separate internet for testing massive firmware upgrades and innovative new approaches is underway with GENI, and creates a great opportunity to building in security and operational integrity.  The technology of OpenFlow, designed to slice up a router to enable researchers access to devices to test ideas without requiring entire new devices or introducing downtime, does cause me to pause and consider the possible inherent risks:

OpenFlow program can be added to almost any router, where it acts like a remote control for the proprietary algorithms and hardware inside.”

This project is highlighted in the article and does have a given amount of inherent risk – introducing such an access vector to core internet routers may create greater interruptions initially then are prevented.  Careful consideration should always be taken when adding features to systems that are inherently single tasked (this is not solely due to the vulnerabilities that may be introduced, but to the increasing degree of complexity added as a result).
Complexity has proven time and again to be the greatest threat to technology, so any increase should be done consciously and expertly to ensure that the entire control environment reflects these changes.

Creating a crashproof internet is an important effort (especially considering the impacts of Michael Jackson on social networking sites and Twitter with Iranian elections), but one must remember the internet is a service provider and as such contingency plans must be devised.  Separate network connections, satellite, and off-the-internet (OII) processing must exist.  Consider how your business would be affected without the internet; with a loss of half the planet; with a loss of consistency in uptime.

Preparation is great business and a necessary control safeguard advised by numerous regulations.

Best regards,

James DeLuccia IV