There has been a lot of talk about container data centers lately (sometimes also called modular data centers). Most of the time we are only presented with an image of a branded shipping container, but let’s face it, all the interesting stuff is on the inside!

So, since we were curious, we here at Pingdom dug through various product pages and presentations to find pictures of the insides of container data centers. As you will see, different manufacturers have very different approaches to building these things.

We have included pictures from Rackable’s ICE Cube 40-foot container data center, and Sun’s 20-foot Project Blackbox (recently redubbed to the slightly less appealing “Sun Modular Datacenter S20”), along with some basic specs for those interested.

Project Blackbox

• 20’ x 8’ shipping container
• 8 racks (7 racks with a total of 280 rack units available for servers)

The ICE Cube

• 40’ x 8’ shipping container (there is also a 20’ option)
• 28 racks (with 1,400 rack units available for Rackable’s half-depth servers)

It should be mentioned that there are also other options for container data centers, such as the FOREST data center from Verari (which probably looks entirely different inside as well).

Judging by recent news, it looks like container data centers aren’t just a fad, but have come to stay. Sun has been pushing Project Blackbox a lot, and Rackable has stated that they expect to ship 20-50 container units in 2008. Microsoft will be using between 150-220 container units to build up their Chicago data center, possibly from several different providers. Perhaps others will follow suit?

Images courtesy of Sun and Rackable.

Tagged with:data-center, Photos, pictures, rackable, servers, sun

There is no denying that there are some significant cultural differences between East and West. Since we here at Pingdom deal a lot with web hosting companies and are generally interested in the hosting industry, we have on occasion stumbled upon some Asian hosting websites which to our Western eyes have a very different look (and not just because we can’t read a word ).

This post has screenshots of some interesting examples from Japan, China and South Korea. The designs are sometimes strikingly different from what you would expect to see from a Western hosting company website, and often quite charming.

Colors and cartoons

Keep it colorful… very colorful. And don’t forget the happy cartoon characters (leprechauns, anyone?)

Screenshot from www.lolipop.jp. (Japan)
—

Screenshot from www.muumuu-domain.com. (Japan)
—

Screenshot from www.kornet.net. (South Korea)

Cram in a LOT of information

We don’t understand a word of this, but that looks like a lot of information crammed into one page…

Screenshot from www.mireene.com. (South Korea)
—

Screenshot from www.firstserver.ne.jp. (Japan)

Cute animals

Cute animals and puppies seem to work well all over the world, including China and Japan.

Screenshot from www.winet.cn. (China)
—

Screenshot from rose.ruru.ne.jp. (Japan)

Some other general observations

• For us in Europe at least, websites in China, Japan, South Korea and other Asian countries tend to load really slowly.
• You basically can’t find a Chinese hosting website that doesn’t offer some form of live support chat.
• Did we mention that we really don’t understand a word on these websites?

And before anyone makes this comment: There are of course plenty of examples where the hosting website looks more or less like a conventional Western website. But that wouldn’t have been much fun to look at, would it?

Tagged with:asia, china, japan, screenshots, south korea, web-design, web-hosting, website

We started using FeedBurner just a couple of months ago. This has allowed us to get some reasonable statistics on the number of people who have subscribed to the Royal Pingdom RSS feed.

We are happy to say that the numbers have been growing, slowly but steadily. Lately, our blog has started to go above 1,000 feed subscribers (readers). The record so far was on April 29, when we had 1027 feed readers.

Admittedly, the numbers from FeedBurner are approximations, but the trend is clear: more and more people are subscribing to our blog feed, which is great. The number of feed readers have increased about 60% on average since mid-February.

Thanks for reading this blog. We promise that will do our best to keep publishing interesting content for you.

(And if you’re reading this from our RSS feed, thanks for subscribing! )

Tagged with:blog, feed, feedburner, rss, statistics

This survey shows how much 16 of the largest and most popular social network sites have been unavailable during the first four months of 2008. How much has MySpace, Facebook, Friendster, Twitter, LiveJournal and many others been offline? Read on to find out.

The monitoring for this survey was done using the Pingdom uptime monitoring service.

We have included two tables, one with a summary of the downtime and uptime for the entire period, and another one which breaks down these numbers month by month.

Downtime summary, January through April

This table is sorted so that the highest downtime is at the top. In other words, it is one of the few lists where it is a good thing to be near the bottom.

* Pownce has only been monitored since January 22, when the site went public.

MySpace comes out on top, at least when it comes to website availability. Their website has only been unavailable for little over an hour so far this year. At the other end of the spectrum we find the Twitter website, which has been unavailable for more than 37 hours (a day and a half).

Orkut, Facebook, Classmates.com, Yahoo! 360, LiveJournal, Xanga and MySpace all have a 99.9% uptime or better so far in 2008, which has to be considered very good.

We also added Pownce for monitoring on January 22, when the site went public. Pownce is Kevin Rose’s post-Digg social networking site. (The other founders are Leah Culver, Daniel Burka and Shawn Allen.)

Downtime month by month

Since we thought it would also be interesting to have a look at the downtime numbers on a monthly basis, we have included a table for that as well. We have sorted the sites just as above, so the largest total downtime is at the top.

* Pownce has only been monitored since January 22.

This monthly view shows us that the majority of Twitter’s website problems were in January and February (it still had more than seven hours of downtime in April, though). Of course, Twitter also relies heavily on the use of SMS, but we did not monitor that aspect, only that the website responded (just as we did for all the other sites in this survey).

Friendster had the vast majority of its downtime in January, and has since then been very stable.

Bebo has, after a rocky January and February, stabilized their site significantly. AOL announced a deal with Bebo in March that it will acquire the social network for $850 million.

As we have mentioned in the past, all websites have downtime sooner or later. Social networks, by their very nature, will have a lot of visitors who return to the site frequently, and a lot of page views per visitor, so any downtime they have tends to be highly visible to their users. With many of these social networks counting their user base in the millions, it is a significant technical challenge for these sites to keep downtime to a minimum.

A note about the monitoring: All monitoring was done using Pingdom’s uptime monitoring service. If a web page is not reachable, returns an error, or takes longer than 30 seconds to load, it is considered as down. Downtime is always confirmed from two geographically separate locations.

Tagged with:2008, bebo, classmates.com, downtime, facebook, Friendster, hi5, last.fm, linkedin, LiveJournal, monitoring, myspace, orkut, pingdom, pownce, reunion.com, social-media, social-network, survey, twitter, uptime, uptime-monitoring, website, windows-live-spaces, xanga, yahoo-360

The Swedish magazine Internetworld, part of the global IT news service IDG, publish a big web hosting survey once a year where the 15 largest web hosting companies in Sweden are evaluated based on uptime and quality of service. Two years running, we have helped IDG design and perform this test, handling all monitoring tasks for IDG with the Pingdom uptime monitoring service.

This is the only test of its kind in Sweden aimed at companies and web entrepreneurs and is a big, annual media event discussed both inside and outside the hosting industry.

This year was the most detailed and fair survey yet. We monitored all web servers and mail servers belonging to the web hosting companies in the survey for a period of three months. All in all, more than 1,200 tests were performed every minute during this period, only for this survey.

We even ended up involving the tested web hosting companies themselves by creating a special control panel for them where they could add or remove servers for monitoring from a list of servers assembled by us, thus giving them a chance to exclude any servers that were not in their production environment (we of course verified the information).

Above: A small snippet of the summary table in the web hosting survey article.

On top of the very thorough uptime monitoring performed by Pingdom, IDG also tested other aspects such as customer service, functionality and ease of use.

Internetworld and IDG.se published the survey in March this year. You can read the whole thing in PDf format on their website. Note that all text is in Swedish.

Tagged with:idg, Internetworld, magazine, monitoring, pingdom, survey, sweden, uptime-monitoring, web-hosting

You don’t hear much about the state of the internet in Africa, so we here at Pingdom decided to find out how Africa’s internet is doing. We looked at data traffic and internet penetration (how many in the population have internet access), and came up with some very interesting numbers.

Bandwidth (traffic)

Internet exchange points (IXPs) are locations where internet service providers (ISPs) interconnect so that they can send traffic directly to each other instead of having this traffic routed via other providers (and potentially very long distances), cutting traffic costs and increasing performance in the process.

The IXP throughput can give us a good idea of how much internet traffic is being handled in Africa.

As you can see, the IXPs in Africa count their throughput in megabit per second, and sometimes even kilobit per second. The rest of the world tends to be concerned with gigabit per second.

To put the above numbers in perspective, here are the traffic numbers for some large IXPs outside Africa. We will let the data speak for itself here:

Internet penetration

The average internet penetration among the population in Africa is 4.7%. The world total is 20% (Europe and North America have 43.4% and 71.1% respectively). But even if the internet penetration is much lower in Africa compared to the rest of the world, it is nowhere near the order of magnitude that is being indicated by the IXP traffic.

Even though 4.7% may sound like a really small number, it is the equivalent of over 44 million people with internet access.

Broadband penetration, however, is much lower at a mere 0.1%. There were only 1,097,200 broadband subscribers in the whole of Africa as of September 2007. Most likely broadband connections are restricted mostly to major companies, schools and authorities.

Conclusion

Pingdom is based in Sweden, so we are spoiled with very good internet connections, and we take them for granted (as do many of you readers as well, we are sure).

Take a close look at the numbers in this post, because they really show how far behind Africa is on the internet. Hopefully significant progress will be made since the internet has become a vital part of the global economy. Africa will fall behind even more in the long run if they are not able to participate on the internet to the extent and capacity of the rest of the world.

Data sources:

We merged data from two different sources to get a relatively comprehensive list of African IXPs and their traffic: Association of African Internet Service Provider Associations, and the IXP directory at Packet Clearing House. We took the highest reported traffic values we could find when we merged the data.

The internet penetration data is from Internet World Stats.

Tagged with:africa, bandwidth, broadband, internet, isp, ixp, traffic

For more than five hours today the so-called Slashdot effect was completely absent on the internet. Why? Because Slashdot.org, the extremely popular technology news website that inspired Kevin Rose to create Digg, was according to our monitoring unavailable from 09:38 CET this morning (03:38 US EST) and only came back up a few minutes ago (14:48 CET, 08:48 US EST).

Sourceforge (Sourceforge.net), which owns Slashdot, was also unreachable.

The Register should have credit for reporting first on this (they even contacted us here at Pingdom for a quote).

Tagged with:downtime, slashdot, sourceforge, The-Register, uptime, website

Google Search has been localized to a lot of languages. And with a lot, we mean A LOT. Aside from languages like Swahili, Tamil, Twi, Scots Gaelic, Gujarati, Sesotho, Uighur and many more, Google Search has taken localization one step further by even including fictional languages:

Leetspeak

Google Search in Leetspeak.

Klingon

Google Search in Klingon.

Bork, bork, bork! (Swedish Chef, Pingdom’s favorite Muppet)

Google Search in Bork, bork bork!

Pig Latin

Google Search in Pig Latin.

Elmer Fudd

Google Search in Elmer Fudd.

Tagged with:elmer-fudd, fun, google, hacker, humor, klingon, languages, leetspeak, localization, muppet-show, pig-latin, search, star-trek, swedish-chef, Tips and tricks

The first hacker attack ever to cause physical harm to its victims may have taken place this year. It happened over the Easter weekend, but barely got mentioned around the web, which is surprising. Maybe it happened too close to April Fool’s Day for journalists to believe it was for real.

Here’s what happened:

Hackers planted flashing animations that triggered headaches, seizures and lockups for visitors to an online support forum for epileptics. The attackers posted hundreds of flashing gif images, and also injected Javascript into some posts that redirected the visitor’s browser to a page specifically designed to trigger seizures in epileptics.

Similar things have happened before, but unintentionally. Maybe some remember the incident in Japan back in 1997, when an animated Pokemon cartoon caused seizures in epileptic kids watching the show on TV. Since then, animated movies and video games often have disclaimers that they may contain flashing images and content that can cause seizures.

Fortunately not all epileptics suffer from this problem. Only 3-5 percent of epileptics are photo sensitive, meaning they can experience seizures from flashing images or certain patterns. However, with approximately 3 million epileptics in the US, that is still up to 150,000 people in the US alone.

You can read more over at Wired, which seems to be one of the few that covered the forum incident, and also the resulting press release from the nonprofit Epilepsy Foundation, which hosts the attacked forum.

Tagged with:epilepsy, forum, General, hacker

We take search engines for granted these days. They exist because they have to. Without them, going through and finding relevant information in the millions and millions of pages that exist on the web would be an almost impossible task.

For example, it would have been extremely time-consuming for us here at Pingdom to put together this blog entry if we couldn’t easily search the web for the information we needed.

Above: The big three. Or…?

Here are nine things you probably didn’t know about search engines.

1. Invented in 1936?

The basic idea that would lead to the invention of hypertext and the reasoning behind the need for quick information retrieval of such stored information (the equivalent of today’s search engines) was published way back in 1945 by Vannevar Bush, an American engineer and science administrator. The essay, As We May Think, may have been written as early as 1936. His concept of a “memory extender” device, a memex, contains pioneering thoughts that ultimately led to the creation of the WWW.

2. A magic, automatic retriever of text

The first actual search engine was created in the 1960s by Gerard Salton. He and his team at Cornell University created the “SMART information retrieval system” (SMART stood for Salton’s Magic Automatic Retriever of Text). Gerard Salton is considered the father of modern search technology.

3. First on the internet

The first internet search engine was called Archie, made to index FTP archives. The name Archie is just Archive with the “v” removed.

4. First on the WWW

The first web search engine was called Wandex. It was released in 1993 and used an index created by the first web crawler, World Wide Web Wanderer, written in Perl by Matthew Gray at MIT. Matthew Gray now works at Google.

5. Not much to do

In December of 1993 there were only 623 websites on the internet, which made work for the first web search engines a lot easier than today, when there are more than 162 million websites on the internet.

6. First full text search

WebCrawler was the first search engine that indexed entire pages, and therefore the first to provide full text search like today’s search engines. It launched in 1994. Before then the search engines had only indexed page titles and header information. Today WebCrawler has transformed into a meta search engine, using results from Google, Yahoo, Live search, and others.

7. The Proto-Google

The technology that would become Google was originally called BackRub, a project Larry Page and Sergey Brin started working on in 1996.

8. Yahoo and Microsoft late in the game

Yahoo and Microsoft didn’t have their own search engine technology until 2004. Yahoo Search used data from AltaVista and Inktomi, and was even powered by Google for a while. Microsoft’s MSN Search (now Live Search) followed a similar path (but didn’t use Google) and didn’t launch their own technology until early in 2005 (beta in 2004).

9. The big three - NOT

Google, Yahoo and Microsoft are usually considered the “big three” in search, but this is actually incorrect, at least if you consider the total number of searches world wide. Baidu, the Chinese search engine, surpasses Microsoft’s Live Search in the number of searches made… which would make Google, Yahoo and Baidu the big three.

Tagged with:altavista, archie, baidu, Engineering, google, history, inktomi, internet, live-search, microsoft, search, search-engine, wandex, web, web-search, www, yahoo