The year 2008 has come to a close. Its time for blogs to tally the successes and failures of web sites and their services. It is another year of product launches and shaky web service consumer confidence. The list of failed launches is always quite high. Inadequate testing and poor anticipation of demand lead to the always popular server flood on opening day. The idea that you have an overabundance of customers may be enticing, but they may quickly becoming an overabundance of dissatisfied customers. Not all products are too popular. Some products simply fail at the gate. They tested great in the isolation of the development group, but when integrated with unfamiliar systems and add some traffic to shake the joints and everything comes crumbling down.
Unfortunately, people do not line up outside the web site for the midnight release. Web consumers will try to cram themselves through the door allowing essentially no one inside or just crash the service completely. For this reason, we make our New Year’s resolution once again. This year we will anticipate Internet demand and consider methods to mitigate high demand time windows. This means no more failed launches, no more unavailable updates, and no more unreadable high-profile news releases.
Many sites now employ a static page during times of high demand or DDoS attacks. The reduced burden of this landing page allow users to quickly retrieve the desired resources and remove themselves without creating large load, much like a busy clerk gets you in and out without the chit-chat. This is similar to hosting high-bandwidth and static content on Amazon’s S3 to prevent network bottlenecks. Of course, the system must still be tested for its capability to handle the processing load of dynamically generated content.
Prior to the economic slowdown, we were hearing some promising news of the network outages or “growing pains” of Mozilla, Facebook, and PirateBay. If we’re lucky, then we’ll continue to see more outages due to growth in the coming months. These products/services will remain popular for sometime to come and their user bases will continue to grow as the number of Internet savvy users continues to grow as well. Does the cost of unused capacity cost more than the loss of business and customer confidence during high capacity hours? The cost is definitely acceptable if your business is focused on up-time and networking solutions. However, the passive blog and local services industry may not be so easily convinced. Consider the issue in terms of an actual commercial storefront. Your customer finds your business center and requests service. If there is no service, then the customer will not wait for it. The customer can quickly find an alternate solution in the Internet marketplace.
Storage and bandwidth are cheap. There are a variety of options for storing and hosting large amounts of static data as well as many options for beginning to create a distributed load-balancing database. It is poor choice to wait for your system to become the victim of traffic fluctuations and lose consumer confidence. Load test your software in the production environment and start scaling UP!
http://royal.pingdom.com/2008/11/11/ddos-attack-crippled-the-bbc-website/
http://royal.pingdom.com/2008/10/14/demand-for-version-30-is-overwhelming-the-openofficeorg-website/
http://www.cbc.ca/world/story/2007/10/30/olympic-ticket-website.html
http://torrentfreak.com/pirate-bay-facing-server-troubles-and-downtime-080729/
http://www.forbes.com/2007/01/30/info-traffic-jams-oped-cx_pk_0131network.html