Earthquake...servers and sites..

 

Gold Member
Username: Lklives

Post Number: 7084
Registered: Jan-06
The Most of the traffic has already been diverted over alternate routes around the broken fibre optic cables, both by us and by our backbone providers (i.e.. Asia Netcom, who own some of the cable capacity). In this way, your connectivity was restored very quickly. However, the alternate routes in most cases are not as fast as the original fibre optic cables (i.e.. Satellites are not as fast as fibre optics - they have additional latency and can have some packet loss, and re-routing via Australia means additional about 50-100ms additional latency as it is further out). So while connectivity has been restored it is not yet optimal. We are continuing to do what we can on our end with additional adjusting of the routes across our backbone providers to further improve the situation.

The actual undersea fibre optic cables are still undergoing repairs, and the backbone providers still have their ships in the ocean around the clock, dragging it up onto their ships and repairing the severed sections. There are multiple ships from multiple carriers doing their respective repairs on their lines, so restoration speed varies. The repairs are taking longer than they originally estimated due to there being multiple parts severed and very bad weather/seas which are impeding their progress. They have not given a firm ETA on all repairs as it also depends on the weather to be completed but are expecting data to start flowing on the cables in a week or so, although absolute full capacity restoration might take longer. China Netcom (which is one of the backbones we have connections to) have said they expect all their cables to be at 100% restored capacity by 15/Jan with partial fibre optic repairs completed this week for improved performance, another backbone has said about a week (8-9/Jan) for their lines. There was considerable over-supply of fibre optic bandwidth prior to the earthquake, so not all cables will be required to restore speed to the previous level, and even if half the cables are restored service should be more or less at the same level already.

In the mean time, you will see that ping times and packet loss are somewhat higher than usual, but it should work.

You can rest assured, that we have negotiated top-level bandwidth and access for your servers and equipment over normal residential or similar users from our backbone providers, so you are getting premium access across the remaining bandwidth available on the backbone's alternate routes. Thus, you are receiving the best connectivity and bandwidth from the backbone providers that they have available, so as they improve their connectivity, you will be the first to see the increase in speed and performance.

On the upside, the Taiwan earthquake breakage has lead OFTA (the Hong Kong Telecommunications Authority) to review a emergency disaster co-ordination plan between the various backbone providers, so that in the event of a future massive natural disaster or similar, capacity can be quickly co-ordinated and shared/exchanged between the backbones on agreement. However, a similar large earthquake disaster is said to not occur for at least 100 years so it is very rare to have an earthquake of this magnitude.

You can also read more on the earthquake, OFTA, the cables, and current progress at the following sites, but note that not all sites are reporting accurately as different companies have different repair schedules and plans:

http://www.expressindia.com/fullstory.php?newsid=78964

5 ships are now on-site repairing the severed submarine fibre optic cables.

The exact breakage's in the cables have been identified:

Four undersea cables linking Taiwan to south-east Asia, the United States and Europe have a total of nine ruptures caused by the December 26 earthquake off Taiwan's south coast.
A cable connecting China to the U.S. Has three ruptures, one connecting south-east Asia to the Middle East and Western Europe has two ruptures, and two cables forming part of the Asia-Pacific Cable Network have 2 ruptures each.

Most of our clients have the majority of connectivity restored, with almost no locations anywhere in the World unable to reach our network (even if slower than usual). Ping times to the USA, while normally 150-200ms, are currently hovering around 200-350ms. There may also be some packet loss depending on the data route taken and the time of day (as all the alternate routes globally are stretched to their limits).

One of our backbone providers China Netcom has repeated that, weather permitting, within the next week to two at least one of the direct fibre optic lines should be repaired to the USA, which will restore routing from alternate routes (i.e.. Around Australia, Europe, satellites) back to using direct HK-USA routing. This should drop ping times to most USA locations back to the usual 150-200ms range, and will indirectly reduce packet loss to all other locations because the alternate routes will have less traffic and less strain, and thus will not be overloaded anymore.

Also note that because most USA/Canada and Western hemisphere traffic is currently going via alternate routes such as Australia or Europe, the speed from a USA computer to a server with us in Hong Kong also depends on who the USA ISP peers with. Some cheaper or less well connected USA ISPs may not have many peering partners (e.g.. May not peer with a European ISP or similar) and thus our alternate routes may not be reaching that particular USA ISP optimally. In those cases, ping times may be 300-400ms or above.

If you have such a case as above, then send us a traceroute output to us, and we will review it with our backbone providers to see if there is any way to route around your ISP's peer deficiency via alternate routing. It may or may not be possible but we will try our best to assist you.

As mentioned above, once at least one of the direct HK-USA fibre optic cables is repaired, direct communications will be vastly restored, so even USA or overseas ISPs without good alternate connectivity will be restored at that time. In addition, alternate routes are being added and adjusted daily, and with feedback from our clients via traceroutes and similar we are helping to improve the situation further, so please give us feedback if you notice improvements or degradation in a particular route.

Global Marine Systems Limited, the independent market-leading subsea cable installation and maintenance company, is playing a pivotal role in restoring vital telecommunications networks in Asia after an offshore earthquake on 26 December caused the largest outage of telephone and Internet services in recent memory.

From the article:

Measuring 7.2 on the Richter scale, the earthquake damaged the subsea fibre-optic cables which facilitate communications between the mainland, Taiwan, Hong Kong, the USA, Southeast Asia and Europe. Millions of people have experienced severe disruption in accessing website's and international dialing facilities are also affected. Mainland service providers fear it will be mid-January before the damage can be fully repaired.

Global Marine vessels, CS Cable Retriever based in the Philippines and CS Wave Mercury from Shanghai mobilized immediately to Taiwan to support the work. In addition, Global Marine's JV partner SBSS (SB Submarine Systems) based in Shanghai, made the cable ship ship Fu Hai available to carry out repair work in support of the major telco's affected by the disaster.

"The number of simultaneous cable failures was unprecedented and we made every effort to get Global Marine and our partners ships to the region to quickly start repairing the damage and restoring network facilities to South East Asia," comments Global Marine's Asia Director Ian Douglas. "We were particularly pleased that SBSS was able to provide repairs services to key customers even though the contracted maintenance stand-by period had ended."

"The telcos in the region rely on our ability to address cable faults quickly and efficiently. We are fortunate to have available the Fu Hai which has completed its six months duties in the maintenance but is still available during January to continue carrying out these emergency repairs prior to commencing another commercial project." Mr. Douglas continues.

Some of the cables are trapped under the seabed and others are tangled, making repairs exceptionally difficult. The ruptures are more than 3300m below sea level well beyond the reach of the Remotely Operated Vehicles which are commonly used in such projects. Instead, the traditional recovery method of using grapnels are being utilized for this operation; grapnels are used to pick up cables from the seabed by being dragged from side to side across the path of the cable until it can be hooked and lifted to the surface for repair on board.

"But this does highlight the vulnerability of our global communications network and the extent to which we rely on it." said Gabriel Ruhan, CEO of Global Marine. "This was a significant earthquake, and we are glad we had vessels in the area which could so quickly be brought in to assist in the recovery and repair."

In other news, Asia Netcom has done some major re-routing work today, some based on your feedback to us, and we are hoping this has helped to further improve our (and your) network performance. As usual, if you do notice any change in network performance, notice us so we can reflect positive or negative outcomes.

Update: 12/Jan/07

As informed by Asia Netcom last week, this week has seen a drastic improvement in their network response times and almost no packet loss, thus greatly improving our network response time and capacity as well. They informed us they are now close to 90 to 100% capacity.

In addition, we have increased our capacity with Asia Netcom as they were among the first of our backbone providers to restore almost full service, so more traffic is being re-routed via them now (as opposed to REACH and the other backbone providers we use).

All clients are welcome to see the improved speed by going to:
http://www.net.berkeley.edu/cgi-bin/traceroute

And entering your website address or IP address and doing a traceroute directly from the test site at Berkeley in the USA. Ping times should be closer to 220-250ms now, compared with 300ms or more a week ago, and now with virtually no packet loss.

Another site you can do ping and traceroute tests from is:

http://www.network-tools.com And enter your website/IP address.

So for all routes that can be re-routed via Asia Netcom are being re-done now (including a significant part of the USA and Canada), and we will try to move more traffic to Asia Netcom and optimize routing further.

As Asia Netcom cannot handle all global locations and all international traffic alone, some traffic will still travel over the other backbones (i.e.. NWT, REACH, China Netcom, NTT) who may still be relying on slower satellite uplinks and alternate routes via Europe, Japan, or Australia. Thus, to certain locations there may still be slower speed or packet loss, depending on which ISP the other location uses, who that ISP peers and connects with, and so on. If you are still experiencing considerably slower speeds, email us a traceroute from your location and we will see what we can do to re-route or optimize traffic to your location.

You can read more about the satellite uplinks in use at:
http://home.businesswire.com/

And for those who are interested to know which cables were affected by the Taiwan earthquake, we have a nice map with an overlay of the fibre optic cables and their status (as of 11/Jan/07). See the following graphic. Note that Hong Kong is the best connected Asian location where the large majority of the fibre optic cables are coming out of (directly below the words CHINA on the map), and that most international cables go out from Hong Kong and travel between Taiwan and the Philippines:

For the remaining backbone providers apart from Asia Netcom, China Netcom expects significant improvements by the 15-20/Jan/07, NWT informed us "soon" and most likely within the next 1-2 weeks, REACH informs us they should have further improvements by next week, NTT said they are constantly re-routing traffic and should have further improvement within the next few days over the weekend.

Bottom line... This is an act of God and NO ONE has any control over this. All we can really do is wait.
 

Gold Member
Username: Lklives

Post Number: 7250
Registered: Jan-06
Just an update about dsscommunity and its server, for those that use it..

as U most know, the earthquake damaged many communication lines, 2 weeks ago in Taiwain, disrupting many sites, including dsscommunity..

some people have experienced NO problems getting into the site and haven't had any impact...it all depends where U live , and how the server has been routed to your specific location and provider..

others, like myself have seen a noticeably difference...pages open very slowly or hard to even enter, especially at heavy traffic times..McSunny and myself have provided a free proxy link many times here to use for faster pages etc..

latest news is that some of the fiber pipes have been repaired already, and all should be repaired/replaced by end of week or sooner...when this is completed, there will be a HUGE improvement from ever before...server will be able to handle twice the traffic with much more bandwidth....so sometimes some good comes out of the bad..

BTW...I myself have noticed an improvement the past couple days, so be patient, all will be better than ever, very soon..
« Previous Thread Next Thread »



Main Forums

Today's Posts

Forum Help

Follow Us