Weekly internet health check, US and worldwide

ThousandEyes, which tracks internet and cloud traffic, provides Network World with weekly updates on the performance of three categories of service provider: ISP, cloud provider, UCaaS.

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 Page 7
Page 7 of 16

On Oct. 27, Hurricane Electric experienced an outage that affected customers and partners across countries including the US, the UK, Russia, Brazil, France, Hong Kong, New Zealand, Finland, Japan, Germany, the Netherlands, and Canada. The outage was divided into eight occurrences over two hours and 10 minutes. The first, lasting around 5 minutes, was observed at around 9:29 p.m. EDT, centering on Hurricane Electric nodes located in New York, New York, and Chicago, Illinois. Around 15 minutes after it appeared to clear, the second one started. Five minutes into it, nodes in Ashburn, Virginia exhibited outages. Around 10 p.m. EDT, all the nodes appeared to recover, but 10 minutes later the nodes in Chicago began exhibiting outage conditions again for about two minutes. Fifteen minutes later, nodes in New York and Chicago  again exhibited outage conditions. Nodes in those locations repeatedly appeared to recover then exhibited outage conditions for the next five occurrences. The total outage lasted around 40 minutes and was cleared at around 11:40 p.m. EDT

On Oct. 25, Rackspace Technology experienced an outage affecting some customers and partners across countries including the US, France, the UK, Spain, Singapore, Hong Kong, Vietnam, Mexico, Chile, Switzerland, Canada, and the Netherlands. The outage, lasting around 28 minutes, was observed around 10:46 a.m. EDT and appeared to center on Rackspace nodes in Chicago, Illinois. Twenty minutes later, some of the nodes apparently began to recover before clearing at around 11:15 a.m. EDT.

Update Oct. 25

Global outages across all three categories dropped from 387 to 354, a 9% decrease compared to the week prior. In the US, outages decreased from 185 to 154, down 17%.

The number of ISP outages globally dropped from 281 to 254, down 10%, and US outages dropped from 150 to 112, down 25%.

Globally, cloud-provider outages decreased from 30 to 22, down 27%, while in the  US, they were cut in half, from 10 to five.

There was just one collaboration-app network outage and that was in the US. The previous week there were two outages globally.

There were two notable outages during the week.

On Oct. 19, Hurricane Electric, experienced an outage affecting customers and downstream partners across the US, Australia, the United Arab Emirates, Thailand, Malaysia, and Canada. The outage was divided into three occurrences over 34 minutes, starting at 1:51 a.m. EDT, centering on Hurricane Electric nodes in Chicago, Illinois. Around five minutes after the first occurrence appeared to clear, outage conditions appeared at nodes located in Los Angeles, California that lasted around eight minutes. The outage was cleared around 2:25 a.m. EDT.

On Oct. 21, Level 3 Communications experienced an outage affecting multiple downstream partners and customers. The outage lasted around 11 minutes total, divided between two occurrences distributed over a 25-minute period. The first lasted around eight minutes and was observed around 5:30 a.m. EDT and appeared centered on Level 3 nodes in Phoenix, Arizona. It appeared to clear, but 10 minutes later the Phoenix nodes exhibited outage conditions again. The outage was cleared around 5:55 a.m. EDT.

Update Oct. 18

Global outages across all three categories increased from 352 to 387, a 10% increase, and from 184 to 185 in the US.

The number of ISP outages increased from 270 to 281, and decreased in the US from 163 to 150.

Globally, cloud-provider network outages rose from 18 to 30, while in the US, they increased from three to 10.

Collaboration-app network outages worldwide increased from one to two, and stayed the same at one in the US.

There were two notable outages during the week. On October 13, GTT Communications experienced an outage that affected partners and customers across countries including the US, China, the UK, Ireland, India, and Japan. The 44-minute outage was first observed around 11:10 p.m. EDT, initially center on GTT nodes in London, England, and San Jose, California. Twenty-five minutes later nodes in San Jose appeared to recover. The outage was cleared around 11:55 p.m. EDT.

On October 11, Microsoft experienced an outage on their network that affected downstream partners and access to services running on Microsoft environments. The 34-minute outage was first observed around 5:05 p.m. EDT centered on Microsoft nodes in Amsterdam, the Netherlands. Five minutes later, nodes in Frankfurt, Germany, began exhibiting outage conditions, affecting more partners. Around 5:15 p.m. EDT, the nodes located in Frankfurt appeared to recover. The outage was cleared around 5:40 p.m. EDT. Given the duration and timing, relative to the location of the nodes at the center of the outage, it is likely to have been a maintenance exercise.

Update Oct. 11

Outages across all three categories during the past week increased from 323 to 352, up 9%. In the US, outages increased from 161 to 184, a 14% increase.

Globally, ISP outages went from 239 to 270, up13%, while in the US, ISP outages increased from 132 to 163, up 23%.

Cloud provider network outages dropped from 26 to 18, down 31%. In the US, they dopped by half, from six to three.

There was one collaboration-app network outage, and that occurred in the US, dropping the worldwide and US levels from six the week before.

On Oct 4, Facebook ‘s backbone network suffered an outage that disconnected its data centers globally, making Facebook, Instagram, and WhatsApp unavailable to all users for more than seven hours. Initially observed around 11:40 a.m. EDT, the backbone network outage triggered a second issue: Facebook's authoritative name servers, detecting that the network connection was “unhealthy,” stopped advertising routes to Facebook's servers, rendering the services inaccessible. Facebook identified the root cause of the outage as a faulty configuration change and implemented a fix. Around 6:20 p.m. EDT, connectivity to services began to return, and the outage cleared around 6:45 p.m. EDT. A more detailed analysis of the outage can be found here.

On Oct. 7, Telia Carrier, a Tier 1 ISP headquartered in Stockholm, Sweden, experienced an outage affecting customers and downstream partners in the US, Spain, Germany, Austria, the UK, the Philippines, Japan, New Zealand, South Africa, the Czech Republic, Egypt, the Netherlands, India, and Canada. First observed around noon EDT, the outage initially centered on nodes in Ashburn, Virginia, and Sweden. Five minutes later the nodes in Sweden appeared to recover, while nodes in Newark, New Jersey, and London, England, began exhibiting outage conditions, increasing the number of countries and downstream partners affected. All except the London nodes appeared to recover 35 minutes after first being observed. The outage lasted an hour and 14 minutes and was cleared around 1:15 p.m. EDT.

Update Oct. 4

Global outages across all three categories last week decreased from 367 to 323, a 12% dip compared to the week prior. US outages remained the same at 161.

Globally, the number of ISP outages increased from 233 to 239, and in the US they increased from 115 to 132, up 15%.

Cloud-provider network outages worldwide dropped from 41 to 26, down 37%, while in the US they increased from four to six.

Both globally and in the US, collaboration-app network outages decreased by two, from eight to six.

There were two notable outages during the week.

On September 30, Oracle experienced an outage that affected customers and downstream partners interacting with Oracle Cloud services in the US. The outage was observed around 4:01 p.m. EDT and appeared to center on Oracle nodes in Frankfurt, Germany,  and Arlington and Ashburn, Virginia. Five minutes later, all nodes except those in Arlington appeared to recover. The outage lasted 7 minutes and was cleared around 4:10 p.m. EDT.

On September 29, NTT America experienced an outage that affected some customers and downstream partners across the US, Japan, and Hong Kong. It lasted around 18 minutes total, divided between two occurrences over a 25-minute period. The first was observed at 7:10 p.m. EDT and appeared centered on NTT nodes in New York, New York. It lasted four minutes and appeared to clear at around 7:15 p.m. EDT. Five minutes later, the second occurrence, lasting 14 minutes, affected the New York nodes plus nodes in Ashburn, Virginia. The outage to nodes in New York appeared to clear 10 minutes later, and the outage in Ashburn was cleared around 7:35 p.m. EDT.

Update Sept. 27

Global outages across all three categories last week increased from 276 to 367, a 33% increase compared to the week prior. In the US, outages increased from 116 to 161, a 39% increase.

Globally ISP outages increased from 186 to 233, a 25% increase, and in the US they increased from 87 to 115, up 32%.

Cloud-provider network outages jumped from 24 to 41, up 71%, while in the US they increased from three to four.

Collaboration-app network outages increased by eight worldwide and also by eight in the US.

Collaboration-app network outages reached the highest number they have reached this year. Both globally and in the US, they were up by eight, increasing from two to 10 worldwide, and from one to nine in the US.

There were two notable outages this week.

On September 21, Oracle experienced an outage affecting customers and downstream partners in the US, Germany, Finland, Singapore, Hong Kong, Australia, and Thailand. The outage was observed around 10:25 a.m. EDT and appeared to center on Oracle nodes in Frankfurt, Germany. Fifteen minutes later some of the nodes appeared to recover, reducing the affected countries to Finland, Hong Kong, and Australia. The outage lasted 24 minutes and was cleared at around 10:50 a.m. EDT.

On September 20, Comcast Communications experienced an outage that impacted downstream partners and customers in the US, Switzerland, China, and Hong Kong. The 18-minute outage was  observed around 4:50 p.m. EDT and appeared to center on Comcast nodes in Sunnyvale and Los Angeles, California; and Houston, Texas. Five minutes after it started, the disruption expanded to nodes in Santa Clara, California; New York, New York; Richmond, Virginia; Pittsburgh, Pennsylvania; and Denver, Colorado. Around 5:05 p.m. EDT, all nodes except those in New York appeared to recover The outage was cleared around 5:10 p.m. EDT.

Update Sept. 20

Global outages across all three categories during the past week increased from 209 to 276, a 32% increase compared to the week prior. In the US, outages increased from 85 to 116, a 36% increase.

Globally, the number of ISP outages increased from 149 to 186, up 25%, and in the US, they increased from 72 to 87, a 21% increase.

Cloud-provider network outages increased from 20 to 24, while in the US they increased from two to three.

Collaboration-app outages worldwide remained the same at two, and also the same in the US at one.

On September 14, Zayo Group, experienced an outage that affected some of its partners and customers in countries including the US, Hong Kong, Switzerland, and the UK. The outage lasted around 15 minutes, was first observed around 3 p.m. EDT, and appeared to center on Zayo Group nodes located in Philadelphia, Pennsylvania. Ten minutes later some of the node appeared to recover, reducing the number of impacted parties before the outage was cleared around 3:20 p.m. EDT.

On September 13, Microsoft experienced a network outage that impacted downstream partners and access to services running on Microsoft environments. The 29-minute outage was first observed around 1:25 a.m. EDT and appeared to be centered on Microsoft nodes in Des Moines, Iowa. Five minutes later, nodes located in Chicago, Illinois, showed outage conditions, resulting in an increase in the number of impacted partners. Around 1:35 a.m. EDT, nodes in Cleveland, Ohio, also showed outage conditions. Twenty five minutes after the outage was first observed, the nodes in Des Moines and Cleveland appeared to recover about 25 minutes into the outage leaving just the Chicago nodes as the only ones still out.The outage was cleared around 1:55 a.m. EDT. Given the duration and timing relative to the location of the nodes at the center of the outage, it is likely to have been a maintenance exercise.

Update Sept. 13

Global outages across all three categories last week increased by one, from 210 to 209. In the US, they decreased from 90 to 85.

ISP outages grew worldwide from 140 to 149 and grew in the US from 68 to 72.

Globally, the number of cloud-provider network outages stayed the same at  20, while in the US they dropped from seven to two.

Collaboration-app network outages remained the same at two, and in the US the number rose from zero to one.

There were two notable outages.

At 1:45 p.m. EDT on Sept. 7, Amazon’s network experienced a disruption affecting downstream partners and customers in the US, Japan, South Korea, Ireland, and the Philippines. It lasted around 18 minutes distributed across two occurrences over a 35-minute period. The first instance appeared to center on Amazon nodes in Incheon, South Korea, and lasted around nine minutes affecting users in the US, the Philippines, and South Korea. Around 2:10 a.m. EDT, approximately 15 minutes after the first occurrence cleared, a second occurrence was observed, also lasted around 9 minutes, and appeared to again center on Amazon nodes located in Incheon, South Korea. It  affected service in South Korea, Japan, and Ireland. The outage was cleared around 2:20 a.m. EDT.

On Sept. 6, Cogent Communications, experienced a series of outages over a half-hour period that affected downstream providers and customers in multiple countries, including the US, Spain, Greece, Germany, Luxembourg, the Ukraine, and Portugal. It lasted 13 minutes divided between two occurrences over a 25 minute period. The first occurrence was observed around 7:15 p.m. EDT centered on Cogent nodes in Sacramento, California, and Bilbao, Spain, and lasted around  nine minutes. The Cogent environment was stable for 15 minutes then experienced a series of four-minute outages observed on Cogent nodes in Bilbao, Spain. The outage was cleared around 7:45 p.m. EDT.

Updated Sept. 6

Global outages across all three categories last week decreased from 286 to 210, down 26%, with most of the decline coming from improved performance in the US where the total outages dropped from 145 to 90, down 37%.

ISP outages globally dropped from 214 to 140, a 35% decrease. In the US they decreased from 119 to 68, a 43% drop.

Related:
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 Page 7
Page 7 of 16
The 10 most powerful companies in enterprise networking 2022