Add to del.icio.us     Digg this

Sep. 9, 2008

According to uptime monitoring service provider Royal Pingdom, high-volume launches, BIND and DNS server failures, Internet infrastructure failure and defective third-party software were some of the many reasons Web servers went completely offline in the first six months of this year.

According to the online report, "no one is immune to downtime," citing Google, Yahoo, Microsoft and Apple that have each made the list, while also noting that sometimes entire parts of the Internet itself "simply break and isn't available for some period of time, sometimes as much as two to three hours."

IT professionals are reminded of the importance of DNS servers. Case in point: YouTube.

When YouTube fell offline for about two hours because ISP Pakistan Telecom mistakenly claimed their IP address space including YouTube's DNS server IP addresses, proving that a single ISP can inadvertently sabotage entire parts of the Internet, either knowingly or not.

It was then determined that it was an unintentional human error that cause a technician at Pakistan Telecom to cause the DNS service outage, but the event clearly spells out the importance of DNS and the large impact it can have on sites.

Additionally, many high-profile launches in the past few months fell victim to simple scaling issues, including iPhone's Active Major Internet Outages.

Another example: Docs & Spread Sheets presentations of Microsoft's virtualized Photosynth service that was overwhelmed by greater-than-expected demand on June 15.

Internet infrastructure problems also affected users of Amazon's S-3 service when it experienced problems in February and again in July 2008.

Finally, third-party buggy software was a major reason more web sites and services suffered a lot of downtime, proving the age-old adage that a chain is only as strong as its weakest link.

According to Pingdom, Web sites "stacked" with third-party PHP scripts and/or applications "may be the weakest link that breaks that virtual chain," making note of the SiteMeter script to handle blog statistics that sent thousands of blogs offline three weeks ago.

In conclusion, Pingdom states that downtime just simply is inevitable, "however, by learning from the mistakes or from the bad luck of others and being well aware of what can go wrong, it is possible to take a proactive approach and greatly minimize the risks of future downtime, and when it happens, at least keep it a lot shorter."

Add to del.icio.us     Digg this

Source: Royal Pingdom.