One feature used by Cedexis Openmix is a simple liveness check service that complements the availability score provided by Radar measurements. Customers use it to augment Radar availability information with a simple HTTP-based ping to a given URL. Up until today we’ve been calling this feature “Pulse Live”, but it is now named “Sonar” thanks to our ever-creative engineering team. The name is nicely descriptive of what Sonar does and complements Radar quite well.

Along with the name change we’ve made some great improvements to the feature. You can now see the current status and recent history for each of your Sonar-enabled platforms. This is particularly useful when testing Openmix application changes or to understand why a change in traffic distribution has occurred.

As always, please let us know what you think. We will continue to make improvements to Sonar in the near future.

The French election highlighted the importance of the Internet as one of the main ways for retrieving information, debating, exchanging, and sharing through social networks in real-time – and from various locations, using multiple types of devices (laptop, tablet, smartphone, connected TV,…).

Last Sunday, before the election results were official, most of the links being shared on the networks in France led to media websites generating exponential growth of traffic from 5pm until late in the evening, as stated by French IT media.

Technical managers of many news websites experienced a massive flow of traffic for several hours which had prepared the technical managers of many news websites. This was particularly the case for our customers, such as “LeMonde.fr”, that used multiple CDNs (Cotendo and EdgeCast, ed) to ensure that an optimized performance level could be achieved even with a quick ramp-up in traffic.

Tweet by lemonde.fr website’s CTO, Olivier Grange – May 6, 2012
“20h on LeMonde.fr: peak of trafic over 4Gbps for Edgecast” explained our customer on Twitter.

A few minutes after this first tweet, LeMonde.fr website’s CTO said that as much as 900 Mbps of traffic had been served from Cotendo with the help of the multi-provider load-balancing platform Cedexis OpenMix.

Constant Focus on the Best Performance

During the first round of the French presidential election, Cedexis Radar spotted a performance slowdown from one CDN. Hopefully, all websites following a multi-providers strategy experienced the performance drop during a few seconds only, just the time to route traffic to a better performing CDN and deliver a seamless experience to end-user customers.

Another media website – 20minutes.fr – had two people on hand during the two rounds of the election in order to address any potential issue in traffic performance and to run the necessary updates in case of emergency.

“20minutes’s infrastructure was entirely upgraded last year. Our webhosting provider ‘Oxalide’ had a special mechanism running in order to monitor the platform and to quickly address any issue if needed, which eventually happened” states Arnaud Limbourg, CTO in charge of New Media for 20minutes.

He adds that “the day of the first round of the election, we used our origin (hosted by Oxalide) and one CDN. But on May 6th, for the second round of the election, we used 3 CDNs. We made those choices because of technical and operational constraints. On the regular graphic updates delivered by Cedexis, we noticed that changes of CDNs happened, sometimes dramatically, depending on the hour or on the choice of operator.”

20minutes opted for routing by performance: best provider selected by Cedexis OpenMix (green). The graph generated 06/05/2012 – 60min period (7:26 p.m. to 8:26 p.m. Paris time)

During an international event of that matter, any incident that would render a website unavailable or slow down the performance of the web service is simply not an option.

“Cedexis helped us to bring consistency in the delivery time of static elements of the pages by automatically managing any network issue that could happen” testifies Arnaud Limbourg.

His conclusion is that “there is such a high level of complexity in the networks that the use of only one provider does not allow to use it at its best. Nothing is 100% trouble-free and Cedexis is the solution to avoid such problems.”

Graphic of Cedexis OpenMix routing decisions for an ISP (May 6, 2012, 5:23 p.m. to 6:23 p.m. Paris time). Of the three providers (CDN), the best provider is the blue used to deliver content to subscribers of that ISP.

Clearly, issues from some CDNs, or in reverse, from some delivery providers to a CDN, are the reasons why some technical managers chose a mix of different providers. It is not an option anymore, it’s the rule when, in some situations, service delivery reaches a critical level of audience or expands on new markets.

We at Cedexis were obviously all on hand to make sure that the infrastructure was perfectly working, and are proud to have helped all our media customers to sustain huge and sudden traffic ramp-ups in the best possible way in order to let people be informed in a timely fashion. This has also helped to avoid the use of downsized versions of websites which come with significant loss of audience and associated revenues from sponsors and advertisers to the profit of more competitive players.

The first Openmix Quick Start application type is used for static routing where Openmix hands back the same CNAME in all cases. This might seem useless, but it is usually the first step for new Openmix applications since you can add Openmix to your CNAME chain without introducing new logic and thus limit the risk of the change. Only later do you add more complex logic to your application in order to route your traffic.

We’ve also enhanced the workflow for editing and publishing applications to separate the act of changing the configuration from pushing it live into production.

Here is an application in “draft” state with unpublished changes.

And there is the confirmation step when publishing into production.

We have additional quick start application types coming in the next several weeks. As always, let us know if you have ideas for quick start application types that would be useful to you.

It’s a question all IT decision-makers ask themselves. To start answering it, Cedexis proposes a new data visualization interface available on our website.

Every day Cedexis publishes new country-by-country reports on the performance of IT providers (Hosting Companies, CDNs and Clouds) providing valuable information on the SOH (State Of Health) of their Internet services as end-users actually see them.

“This daily reporting tool is using billions of measurements generated by users surfing Cedexis’s customers sites every day. In addition, we have also added a feature to measure the performance, country-by-country, for content owners using multiple Clouds and CDNs” states Marty Kagan, CEO and co-founder of Cedexis.

The performance measurements of each of the cloud/CDN providers is generated by the users themselves. This is achieved through the integration of a probe (JavaScript tag) on a Cedexis customers’ web, mobile or business application, which only becomes active only after a page has been fully loaded so that an end-users experience is not impacted.

A billion tests collected every day, but what next?

Deficient performance has huge consequences for companies generating their revenues online including a lack of online turnover, low conversion rates, reduced brand recognition or a reduction in ROI (Return On Investment) on existing spending can often all be attributed to this deficit.

Cedexis combines these quality and performance measurements with an additional service, « OpenMix », which enables real-time traffic routing changes to maximize the end-user’s experience.  Thanks to this real-time decision making service, end-users are always sent to the best provider for them at that point in time, with content owners comfortable in the knowledge that 100% availability and optimal performance for each and every end-user is a reality.

« Our community-sourced approach to measurement helps content and application owners to better understand web and mobile site performance from their customer’s point of view. We can then detect any quality downgrade, like poor loading times or an access network and partner (Hosting Company, Data Center, CDN’s and Clouds) having a problem and through combining this information with our multi-provider load balancing service, dubbed OpenMix, we can swiftly react to those events as they occur. » clarifies Marty Kagan.

To fine tune their multi-provider strategy, content owners are also encouraged to supply additional inputs into the load-balancing that are relevant to their business. Such metrics include provider costs, commit levels, geo-location, KPI’s (Key Performance Indicators) or even feed-back information from a data centers infrastructure (such as UPS status, server utilization, concurrency and bandwidth utilization) allow the content owner to timely anticipate and react to any infrastructure slow-down or failure in a fast and efficient manner.

In addition, Cedexis continues to develop API’s with its partners and other 3^rd parties allowing the ease of integration of other 3^rd party systems and management architectures into the load-balancer logic, for example, with the likes of New Relic.

 Journalists : More information about us is available in the full Press Release.

A 100% owned subsidiary of the Michelin Group, ViaMichelin designs, develops and markets digital products and services to support the mobility of road users in Europe.

Building on a century of Michelin expertise in the field of travel, this activity which was launched in 2001, addresses both business and the general public. It offers a comprehensive range of services such as hotel booking, restaurants, traffic information, tourism, etc. accessible through multiple media: Internet, mobile phones including iPhones, navigation systems, etc.

The ViaMichelin portal covers today more than 45 countries in Europe and maps 9.1 million kilometers of roads and highways. It is available in 11 different languages.

Year after year, new versions were developed with rich content bringing more and more fame to the ViaMichelin online experience, thus generating bursts in traffic.  The technical teams then started to load balance the traffic more precisely in order to guarantee each and every user an optimal online experience.

The challenge

Sometimes incidents happen for a reason and Gilles Ducarre, CTO of ViaMichelin.com,  happened to have learned about Cedexis.

“We used backbone probes for quite a long time to measure the performance of the site and to make sure that a sufficient quality level was met”, Gilles Ducarre admits.

But when a new version of ViaMichelin was rolled out in 2008 with a Web 2.0 orientation, the strategy had to be revised. When the user needs to calculate his itinerary, tens of requests can be sent out that can seriously affect the user’s response and loading times.

“The new version was including elements running in the user’s browser and not on our servers anymore. So we had no clue about the performance from the user’s side”, the ViaMichelin CTO explains.

Cedexis usage

Gilles Ducarre found Cedexis Radar to be the ideal tool to analyze the real-time performance from the user’s point of view. It is for him a valuable decision taking tool, as he has to know the SOH (State Of Health) at all Internet levels, in order to fine tune the contents delivery strategy, but also to anticipate the move into a new market, making sure that maximum performance will be available before even rolling out the service in a new country.

The decision-maker further indicates that “we have given priority to the usage of our data center located in France but, for some destinations, we had absolutely no clue about the response times. Cedexis brought us the real-time metrics and also allows us to use different providers, an option that we are always taking outside of France in order to guarantee optimal and exceptional comfort and quality in the usage of our service.”

When half of the traffic for one service is load balanced between multiple European countries, choosing a single-provider strategy could be considered as obsolete.

That’s the reason why ViaMichelin is also using Openmix, the multi-provider traffic controller from Cedexis.

“Today, we can mix the usage of our provider at origin (Easynet) and optimize in real time when necessary with various CDNs (Cotendo, Edgecast, CDNetworks) in a cost effective way”, continues Gilles Ducarrre.

The results

Several months after the integration of Cedexis, ViaMichelin stopped using backbone probes as results tended to become irrelevant and the tools inconsistent.

“The real-time performance measurement from the user’s point of view combined with the fastest content delivery from the best provider for the user is, for us, the ultimate service. We are not worried about quality anymore, even if peering can sometimes be bad, because we are always sure to be on the best routing platform”, Gilles Ducarre noticed.

He also reveals: “we, however, use Cedexis data to negotiate with our host in order to ask him to optimize his service with content providers as we can see more than he does, thanks to the Radar metrics. In the end, he also benefits from it because he gets more traffic and we gain from a reserved bandwidth that we try to optimize.”

To summarize, with the usage of Cedexis, ViaMichelin has a real negotiating power at disposal with their providers in order to get the best quality of service at the same level of local competitors. Thanks to multi-sourcing, the site is guaranteed to receive a provider’s choice to permanently benefit from using the best of them all.

This usage flexibility is highly appreciated and offers an optimal online experience for millions of users wishing to quickly compute the itinerary of their next holiday road.

Illustration : Flickr / Thx ComputerHotline

This integration opens up the possibility of using any measure that New Relic offers, such as memory, CPU, database, and Apdex score to add a new level of intelligence to Openmix applications. For example, imagine a scenario in which a data center is reaching a critical load threshold. Openmix can be programmed to direct fewer requests to that data center before the data center starts exhibiting performance problems, based on the information supplied by New Relic.

A couple of months ago we showed off a demonstration of this integration. For the demonstration we crafted a simple Openmix application that chose randomly between two data centers unless one of the data centers reported CPU load above a certain threshold. If the threshold was exceeded then that data center was removed as a possible destination for traffic. In this chart you can see the behavior:

• From 23:15 to 23:27 “New Relic Provider 2″ was decided against because it was exceeding the CPU threshold set in the Openmix application
• From 23:27 to 23:48 both providers were chosen (at random)
• From 23:49 to 00:08 only “New Relic Provider 1″ was used because the other provider was over the CPU threshold

Let us know if you’d like to try out this integration and see the power of New Relic and Openmix together.

You can also check out the New Relic press release.

The historical results are available free to anyone in our Free Data section and can be useful as a vendor selection tool. With Openmix, customers can use these data to connect each web visitor to the best performing provider as measured from that visitor’s network.

But which measurement to choose? HTTP Response time is a measure of  latency. HTTP Throughput measures how long it takes to download a 100KB object. So far, most customers use HTTP Response time, but for customers whose pages include larger objects (versus many smaller objects) Throughput may be the better choice.

To make life easier, we recently added a new template to our Openmix Application Library: “thruput-with-rtt.php”.

This app uses Throughput as its first selection criteria, but then checks to see how close the top 2 performing providers actually were. If they are within 5% (easily configured within the PHP), then we use Response time as a tie breaker. Now you can use both!

This new edition of the « Trophées du Cloud » perfectly illustrated the dynamic and the growth of this market. The « Trophées du Cloud » 2012 award the best in 5 categories for their applications and solutions « in the Cloud ».

During several weeks, the 17 members of the jury chosen within the IT industry have been through 115 candidates files generated from 75 different companies. And Cedexis was awarded the “Best Cloud Start-up” trophy !

“We are particularly happy to receive this unique award and proud to be recognized as the best Cloud company on the French market,” declared Julien Coulon, co-founder of Cedexis.

Julien adds that “this award comforts us in our choices and I first want to think of our teams, customers, partners and families who support us since the beginning of this journey. I warmly thank them all for the trust they have put in us. This award is pushing us forward into our R&D efforts. We will present new services in a few weeks in order to make the traffic control between Clouds even more available and even more performing for all Internet users and mobile users”.

You discovered our company? What do we do exactly…

Cedexis was created by the Cloud and for the Cloud : with Cedexis Radar, content editors can measure the performance of the providers (hosts, CDNs, and Clouds) they are using, or could be using, to delivery/host their content.

The unique feature of the Radar consists of transforming each user into a monitoring probe while he is surfing the web, by integrating a simple Javascript tag activated a few seconds only after a page is fully loaded in the browser. This well-proven methodology gives Cedexis the ability to collect a huge amount of data, consisting of more than a billion measurements a day from 34 000 networks in 230 countries.

The measures are publicly available through an intuitive visualization interface.

When performance is slowing down, or when a provider becomes unavailable, the Cedexis DNS load-balancer, dubbed OpenMix, automatically takes the decision to route traffic on the best available provider.

In few words : OpenMix enables its customers to control multi-platform strategies (multi-CDN / multi-datacenter / multi-Cloud).

And finally, Cedexis customers can integrate personalized routing variables coming from an infinite number of sources, such as server loads, agreed commitments, and even the electric consumption of their data centers.

Following this approach, Cedexis opens up the doors of a true world of quality of service delivered by the providers, and can guarantee to content editors that their web/mobile site will always be performing and 100% available, for final users and for search engines crawlers.

Since its creation in 2009, Cedexis managed to satisfy 250 customers, shares its experience with many key partners and transformed the traditional approach of monitoring, and the costly and complex routing and load balancing, into a unique and disruptive hosted service on the market, resilient, on demand, and highly customizable while being cost effective.

Want to know more about us? Contact us now or create an account in few seconds to integrate our tag Cedexis Radar and measure the performance of your website / mobile.

“Just one word : Thank you”. Julien / Follow JulienCoulon on Twitter

###

A few words about EuroCloud…

EuroCloud France is a professional organization that aims to encourage, to support the organization and encourage market development of cloud computing (SaaS, PaaS and IaaS) in France. Its mission is also to clarify the scope of the market potential and conditions for its growth, to better accompany its growth. EuroCloud France brings together more than one hundred companies and over 800 professionals of Cloud Computing in France. Since 2009, EuroCloud France is part of the federation EuroCloud, the first federation of actors of Cloud Computing in Europe, with 27 local branches (by country) distributed in Europe. EuroCloud Europe is an initiative of EuroCloud France (old brand “ASP Forum”).

Medias / Journalists: You wish to contact our team or follow our news? Visit the press room or on our Twitter feed @Cedexis

Today, several availability zones in Amazon AWS US-East region went down i (US-EAST-1). Here is Cedexis Radar detected in real time…

[First edition of this article : 09:50 UTC]

As of Thu Mar 15 09:46:21 UTC 2012, Cedexis Radar has detected what appears to be an outage at 3 of the 4 availability zones (AZs) in Amazon AWS’ US-East region.

The screen shot is of US-East 1A. Our data suggest that the only AZ currently reachable in US-East is 1C. 1A, 1B and 1D appear unavailable.

UPDATE 1: Total outage was around 19 minutes. See second screen shot below.

UPDATE 2: Amazon has confirmed, here’s the current message on Status dashboard.

2:40 AM PDT We are investigating connectivity issues for EC2 in the US-EAST-1 region.

3:03 AM PDT Between 2:22 AM and 2:43 AM PST internet connectivity was impaired in the US-EAST-1 region. Full connectivity has been restored. The service is operating normally.

… a great reason to look into a multi-vendor strategy!

Another event involving maintenance already generated a rant from customers and IT providers.

In the IT news this week you probably saw the reports of an incident involving Microsoft’s Azure Cloud service.

The IT medias has done a great job of summarizing what happened. Read their full reports here, here or here.

There are remaining questions observers (and cloud users) want to ask : How long did the breakdown last? Is the service provider transparently reporting the incident? Can I get independent data anywhere? How can I avoid being impacted by it should it happen again? Is the cloud reliable?

You won’t be surprised to learn that Cedexis has the answers. Of course, it’s our job to help you get through this kind of unfortunate outage…

How long did the breakdown last? Is the information provided consistent with the RFO (Reason for Outage) released by the provider?

This is what Cedexis Radar users very simply observed. We chose to show the availability of the various Azure Cloud regions over 24 hours.

The resulting diagram shows the impacted Azure regions and the duration of downtime.

This diagram was released today and generated from our free data section:

Is it possible to bypass such outage? Is the cloud reliable?

Yes to both questions. Some of our customers (and therefor millions of final users) haven’t been impacted. How is it possible? By choosing a multi-provider strategy involving Openmix to direct traffic based on slowdowns and outages detected via the Radar and/or customized data feed provided by your own infrastructure.

Did you ask for multi-cloud experts? Here we are