Mouse in the House

Posted by mark on July 1st, 2010

Soccer (also knows as football in the rest of the world) team sites, sports and social media sites are experiencing increased traffic – with big spikes during the match and after scores – due to the 2010 FIFA World Cup competition. We thought it would be interesting to measure the uptimes of the websites associated with the qualifying FIFA World Cup teams, the six Federation websites and FIFA.com.

Today we released a new report with the findings of how the websites associated with the 2010 World Cup performed. We were able to test and measure availability from all over the world, and specifically from South Africa with our newly placed monitoring station in Cape Town.

Coincidence (or maybe not?) that a number of the sites with the poorest uptimes were those of teams that failed to qualify among the final eight teams in the semi-finals. Only 10 sites had 100% uptimes and that includes the team site of The Netherlands!
Hup, Holland, Hup! That’s Go, Holland, Go! for our English speaking friends. Hmmmm….wonder who we’re rooting for???

Click here to read the full report or visit the public status page and view the live status and statistical data.

Posted by stan on June 29th, 2010

We regularly invest in infrastructure that strengthens our ability to watch over online business, and this includes adding to our growing list of worldwide monitoring stations. The addition of new monitoring stations is both strategic as well as client driven. Our most recent monitoring station additions are in Jakarta, Indonesia and Cape Town, South Africa, and bring our worldwide grand total up to 48.

We added Indonesia due to the increased user adoption of social networking sites in that country, whose population is said to be ranked among the top social network users in the world. We added South Africa because we anticipated that the 2010 FIFAWorld Cup would bring an influx of web traffic and server stress from that country – plus, we love football/soccer every bit as much as the next fan! (Not that we’re rooting for anyone specifically, but…Hup, Holland, Hup!

Adding these two new monitoring stations lets us test and monitor how any website is performing from the perspective of a web visitor in each of these specific countries.

We value your feedback. Where do you think the next WatchMouse monitoring stations should be and why? Let us know and perhaps you’ll be reading another blog post from us in the near future announcing the location you suggested!

Posted by mark on June 8th, 2010

Detailed Timing Analysis of Embedded Page Elements now Visualised

Currently the WatchMouse website monitors (of type http and https) retrieve the HTML of a page and measure the resolve time, connect time, processing time and download time. Often this information is sufficient to detect slow page loads.

In some cases, however, it would be great to have a more detailed analysis containing all embedded elements (images, external stylesheets, …). This so-called “Full Page Monitoring” is available through the use of the WatchMouse scripting service.

Today WatchMouse introduces a new feature: Waterfall Analysis Diagrams for Full Page Monitors. The image below is an example of the results for a single Full Page Load check.

Love FireBug?

The loading of a page and its elements is visualized similar to the breakdowns you can get from tools like FireBug. And similar to FireBug you can generate a waterfall diagram for any webpage, allowing you to inspect slow loading elements, verify on server response headers, etc.

Next to that, the new Waterfall Analysis has two additional benefits:

• Continuous monitoring: your webpage is monitored every five minutes, with each check providing a full waterfall analysis, and each being available on demand through the dashboard.
• Worldwide monitoring: your webpage will be monitored from 48 different locations worldwide, providing interesting details on the pageload from each of these locations, not just from your current location.

Waterfall diagrams, all the time, any location

These two characteristics of the WatchMouse Waterfall Analysis (continuous monitoring and monitoring from many locations) can be very useful evaluating page load time in the following ways:

• Page load time changes over time. Peak traffic makes a web site possibly load slower, and site changes (like database modifications, or server architecture changes) can also have impact on the page load time. Page load trends can be followed in the standard performance reports in the WatchMouse console, while an in-depth analysis of a slow monitoring sequence can be performed using the waterfall analysis.
• Page load time depends on the location of the visitor. If your website’s target audience is international, you probably optimize the delivery of the page and page elements for all of your target countries by using a distributed cloud solution or a content delivery network (CDN). The Waterfall Analysis performed from many locations provides insight in which of the page elements are the bottleneck for a fast loading website.

Getting started

If you are a WatchMouse client already and you use our “scripting” monitors, there is nothing you have to do to dive into the Waterfall diagrams. Just open the log viewer and select the monitoring results for any of your scripting monitors: You will see the Waterfall Analysis icons show up. Clicking on the icon will lead you to the Waterfall diagram for that specific check.

If you are not using scripts yet, or if your scripts are not set-up as a “Full Page Load” monitor yet, simply follow these steps:

1. Make sure you have one or more unused script monitors available in your subscription plan. You can always add additional scripting monitors to your subscription.
2. Download the Full Pageload template and unzip it:
3. Create a new monitor in the settings, and select type ‘script‘.
4. Upload the Full Page script template (full_page.jmx) that you just downloaded.
5. Switch to “expert mode” and fill in the host and path fields, e.g.: “www.yoursite.com” and “/index.html”
6. Change any other settings to your taste and finally click ‘save’.
7. Within a few seconds you should see the first monitor results and the waterfall via the log viewer.

Thats it!

How does it work? HAR!

The WatchMouse Waterfall Analysis diagrams are based on a number of open-source technologies which are integrated into the our monitoring infrastructure. The visualization of the waterfall itself is based on Jan Odvarko’s excellent HARviewer. As the name implies, the input format for the visualization is based on the Steve Souders‘ HTTP Archive format which is a generic “format that can be used by HTTP monitoring tools to export collected data”.

In short: Monitoring results from the WatchMouse Full Page Monitors are converted to HAR to be able to visualize it. Each waterfall diagram is also directly available for download as a standard HAR file, enabling you to use it in other tools too.

Keep an eye on the WatchMouse Labs blog for an in-depth post about how we convert our monitoring data to HAR and more.

Share it!

Waterfall Analysis diagram can easily be shared with your colleagues, without requiring direct access to your WatchMouse account. To facilitate this, all Waterfall diagrams include a sharable URL, for example:

http://www.watchmouse.com/en/waterfall.php?mid=3&rid=51138&lid=31991975&key=fc37ac9b0a0ba88.

Go ahead and click it!

What do you think?

The Waterfall Analysis diagrams went live today, and are not perfect yet. A few known issues are:

• It won’t work in Internet Explore 6, and only limited support is provided for Internet Explore 7.
• Not all timing parameters are available yet (e.g. resolve time). We are looking into adding these in the near future.
• The waterfall view doesn’t incorporate typical browser behavior (like parallel connections, pipelining, etc.) yet.

What do you think? Is this new feature useful for you? Please let us know what you think by leaving a comment, how you are using it, and what improvements you would like to see.

Posted by admin on May 4th, 2010

We’re excited to announce a new product today called GeoBrand that monitors and detects brand abuse in PPC advertising on major search engines (Google, Yahoo! and Bing) in 30 different countries.

We tested GeoBrand using Interbrand’s “2009 Best Global Brands” list, and found that 81% of the top 100 global brands are the victims of search engine advertising brand abuse including BMW, IBM, Cartier Coca-Cola, Goldman Sachs and L’Oréal. We also discovered through our testing that one of the sectors with the most widespread and blatant brand abuse is the automobile industry.

Search engine advertising programs offered by Yahoo!’s Sponsored Search, Google’s AdWords, and Microsoft Bing’s Search Advertising service allow any business to purchase keywords to display web advertisements on the search engine results page. The recent court case, involving Google and Louis Vuitton (LVMH) regarding use of the LVMH trademark and brand in association with pay-per-click advertising by those other than LVMH, was a highly publicized case and highlighted this problem.

Our research has proven that brand abuse on PPC ads is rampant, and Louis Vuitton (LVMH) is not the only global brand that has an issue. The UK court’s decision however, left the challenge of diligent tracking and handling this type of brand abuse with brand owners. This has been an arduous and difficult task to date considering that each of the major search engines have varying rules and guidelines and that search results differ based on the searcher’s IP address/country, time of day, number of advertisers and a host of other variables.

GeoBrand is an easy to use tool for CMOs, brand stewards, intellectual property and trademark attorneys to continuously and proactively monitor and manage all of their brands and trademarks on the World Wide Web.

The GeoBrand Solution offers:

• Actual tests performed in 30 countries
• Smart filtering technology delivers the most relevant brand abuse cases
• Clean, web-based console for easy management and reporting
• Continuously and proactively monitor and manage all brands and trademarks

GeoBrand is available for professional brand managers, intellectual property and trademark professionals.  Discounts are available for International Trademark Association (INTA) members.

Posted by mark on April 19th, 2010

Traceroute is a computer network tool used to show the route taken by packets across an IP network. Traceroute is often used for network troubleshooting.

Performing a traceroute (or tracert) from your local computer can be very useful, but many times you need to see traces from different locations to gain insight into possible network issues.

As of today, our free online traceroute tool can be used to trace from 40 locations worldwide: http://www.watchmouse.com/traceroute.php.

You can see the tool in action in the screenshot below with a trace from Nagano, Japan to our central server at www.watchmouse.com.

Traceroute is already used in our Root Cause Analysis functionality. When an error occurs, two traces are instantly executed from the used monitoring stations.

We hope this is useful for you and welcome your feedback!

Posted by mark on April 13th, 2010

We enjoy working with a number of leading companies around the world, and thought we’d tell you about a few companies that we recently started working with including:

• Dutch Railways, the principal passenger railway operator in The Netherlands
• 9292OV, the leading expert on travel information and travel planning for Dutch public transportation
• Nuon, a leading Dutch energy company
• Fiat, an international automobile maker
• Zappos, an innovative online retailer

Each of these companies has highly trafficked consumer-facing websites that need to be up 100% and operational 365 days of the year. They saw our product as the best way to monitor the performance of their sites globally.

We welcome these companies to our rapidly growing list of WatchMouse customers around the globe!

Posted by mark on March 16th, 2010

URL shorteners. We use them. You use them. Lots of people use them. URL shorteners like bit.ly are widely used nowadays, but are they really as good as they appear to be?

Mouse in the House digs a bit deeper into the pros and cons of URL shorteners.

On the positive side:

• URL shorteners obviously provide useful features like making a long URL shorter (i.e. so it fits easily in a Twitter message)
• They enable you to track and analyze clicks on a specific short URL
• Some URL shorteners like twt.tl also provide some browsing safety by analyzing the target URL for harmful website code or phishing attempts

But on the negative side, URL shorteners also introduce:

• An additional single point of failure: when a URL shortener service is down (or corrupt) the link won’t work
• Additional load time for a page to fully load

WatchMouse monitored the most popular URL shorteners for one month to find out how they are doing in terms of availability and speed. During that time we monitored 14 URL shorteners and collected the uptime and performance statistics. The uptime results are shown in the chart below:

Uptime is still clearly an issue for some of the URL shorteners. This is important because it has a direct impact on the uptime availability of the website the URL shortener actually directs to. Only goo.gl and twt.tl score a perfect 100%.

The performance results can be seen in the chart below:

According to our data, Facebook’s fb.me is by far the slowest. It adds over two seconds on average to the page load time after the click on a link.  And, quite a few others still take over half a second of the page load time, which is really way too much for just a URL redirection. This substantially affects the user experience.

Another interesting thing we noticed is that only a few of the URL shorteners optimized their name servers (DNS) for international use – i.e. it takes half a second for some of the URL shorteners just to lookup the IP address that is needed for a browser to retrieve a web page. That means, that while it might be fast for a visitor from the US, a visitor from Asia might get some extra waiting time when using snurl.com, for example.

And, while bitly.pro might offer more options than the free bit.ly (like having your own domain name), the paid version is also slower on average than its free counterpart.

Some details about how we measured all this: The URL shorteners were checked every five minutes from one of the 44 WatchMouse global monitoring stations. For each short URL, only the redirection was measured, not the actual loading of the target page. The redirection was expected to be done within eight seconds without any errors (like when a server error occurred or if the expected target URL location was not found in the http header). If that time was exceeded or a second error was established, WatchMouse verified the results using another of its monitoring stations and the result was counted as either poor availability or unavailable.

We plan to continue to monitor URL shorteners and as of today, plan to share the results publicly through our brand new website portal: http://url-shorteners.public-website-status.com/. Here’s a screenshot:

You can also receive Twitter alerts so you know immediately when URL shorteners go down by following http://twitter.com/url_shorteners.

Now it’s your turn to tell us what you think.  Are URL shorteners useful or can you live without them? Does the additional time to load a page concern you?

Update: as pointed out at in some comments here and at other blogs that followed up on this post: The results of bitly.pro shortheners like tcrn.ch is affected greatly by the name servers of the bitly.pro clients, and those servers are not controlled by the bit.ly people. In the performance chart above the resolve time of tcrn.ch is indeed responsible for the lower overall performance. Binged.it, another bitly.pro client is actually faster, due to a great worldwide DNS performance we saw from our 45 monitoring stations in 26 countries. Also, the bitly.Pro service is actually free!

Posted by admin on March 10th, 2010

Welcome to our new blog, Mouse in the House. We’ve been busy these past few months improving our WatchMouse web site with a new home page, navigation and content. As part of those improvements and in addition to our Labs blog which is geared toward more technical types, we’ve added Mouse in the House, a general blog for news and information about WatchMouse, what we’re working on, industry insight and opinion.

We’ll also make company product announcements here so you will be the first to know when we release new products, or make product enhancements and improvements.

Today we released several reports that measured the availability and uptime of the companies’ websites that make up the different Stock Exchange indexes in 11 countries (NASDAQ website status, FTSE 100 website status, CAC 40 website status, OMX Stockholm 30 website status, DAX 30 website status, FTSE MIB website status, IBEX 35 website status, AEX website status, BEL 20 website status, ASX 50 website status, SMI website status). The websites we monitored were of the largest public companies in the world…and a lot of them didn’t fair too well.

We thought it would be fun to stack all the exchanges against each other to see which country’s companies had the best uptimes. Drum roll please…Sweden’s OMX 30 companies had the best aggregated uptime with 99.42%, the United State’s NASDAQ companies came in second with 99.4%, and France’s CAC 40 companies in third with 99.19%. The worst was Australia’s ASX 50 companies with 97.52%.

Stay tuned for more news, information and opinion and watch out….there’s a Mouse in the House!