By Joseph Menn
SAN FRANCISCO, Oct 30 (Reuters) – Despite outages at a few
well-known web sites and ripple effects that occasionally slowed
communications around the country, the Internet came through the
massive storm that swamped New York and New Jersey with
relatively minor problems.
Built for resiliency and buttressed by the adoption of cloud
computing, the Internet functioned largely as it was supposed
to, industry experts said, routing around major disruptions in
one of its central network locations, New York City.
“You don’t hear about big content providers going offline
anymore,” said Jeffrey Young, a spokesman for major delivery
network Akamai Technologies Inc, which has servers
spread among some 1,100 communications set-ups. “We can route
around issues that are occurring.”
New York has two major exchange centers where U.S. backbone
telecommunication providers meet data from undersea cables, said
Doug Madory, senior analyst at Renesys Corp, which monitors
Internet response times. A number of network addresses were
inaccessible on Tuesday afternoon, including many in the New
York area, and connection times for others were slower than
normal, he said, but the disruptions were limited.
Still, a handful of popular web sites, including Google’s
YouTube, AOL Inc’s Huffington Post, and the
network of sites owned by Gawker Media, did experience outages.
Social news site BuzzFeed and News Corp’s financial
site MarketWatch were also reduced to bare-bones versions as
they regrouped on Tuesday.
At least some of the problems were the direct result of data
centers losing power and being unable to fuel their own
generators because of flooding.
Most web sites use commercial data centers rather than
running their own computer servers, in part to ensure security
and stability in emergencies. Many of those data centers offer
back-up services elsewhere.
But New York data centers, including Internap and Datagram,
went down due to the power and flooding problems. BuzzFeed,
Huffington Post and Gawker all crashed because of Datagram.
“How dumb to locate datacenter in a flood zone. And how dumb
to host Gawker servers there,” Gawker founder Nick Denton wrote
on Twitter. The company moved to blog platform Tumblr for a
while, one of a number of creative workarounds made easier by
more advanced web offerings. BuzzFeed moved everything onto
Akamai and Amazon’s web services arm.
An AOL spokeswoman said after Huffington Post’s main data
center went under that it had shifted to a backup data system in
Newark. That worked until all three telecom firms serving that
location went down.
“At approximately 3:30 a.m., network connectivity failed at
the backup datacenter when all three of its providers each
separately failed,” said AOL’s Erin Kurtz.
Huffington Post switched to a skeletal blog platform until
it recovered at eight hours later.
Dow Jones and Datagram could not be reached immediately for
comment. YouTube refused to say why it became unavailable, but
Google’s New York headquarters was closed by the storm. The
company said the video site was available, if slow, for most
users by Tuesday evening.
Many large and mid-sized companies adopted disaster planning
for data after the Sept. 11, 2001, attacks, making sure to have
duplicates of core data in different locations. Smaller firms
have taken the same route by moving to cloud computing, which
generally spreads data across multiple facilities.
“The whole point of the cloud is that companies are
insulated from outages really of any sort, absent a giant
nuclear disaster,” said Bernard Golden, vice president at
enStratus Networks, a cloud software company. “You want your
provider to have facilities in disparate-enough locations so
that even if you have problems in a particular region, your
service is still available.”
