Gmail : la panne, expliquée et les clients, dédommagés !

Google Mots clés : google down, SLA, panne, gmail

Par Louis naugés :

Dans mon dernier texte,qui parlait de la “monstrueuse panne” de Gmail, il y a deux jours, je terminais en espérant avoir des explications de Google.

De nombreux commentaires posaient des questions sur le SLA, contrat par lequel Google garantit un niveau de service de 99,9 %, et de la manière dont Google allait le mettre en pratique.

J’ai obtenu des réponses à ces deux questions.

Panne Gmail : les explications

Deux jours après cette panne, Google a envoyé un message à tous ses clients et publié sur l’« official gmail blog” une explication détaillée de l’origine de l’incident.

Résumons en quelques lignes ces explications :
- C’est un “bug” logiciel (désolé pour les partisans de la grande conspiration mondiale contre Google ou d’une attaque massive contre ses serveurs).
- Google a réalisé une maintenance préventive de l’un de ces centres de calcul en Europe. Une modification récente du logiciel destiné à faciliter la localisation des données au plus près des utilisateurs a déclenché une surcharge d’un centre de calcul. Cette surcharge c’est ensuite propagée à d’autres centres de calcul
- Les ingénieurs de Google ont mis une heure pour trouver la cause de la panne ; ils ont ensuite pu remettre, en deux heures, en état de fonctionnement tous les centres de calcul concernés.
- Aucune donnée, aucun message n’ont été perdus.

Sur ce thème, Phil Wainewright émet une hypothèse intéressante ; Phil est un blogueur célèbre de la galaxie ZDNet, spécialiste du Cloud Computing et des solutions SaaS ; il sera d’ailleurs à Paris le 5 mars pendant les Etats Généraux du SaaS/ASP, où j’interviens également.

Selon lui, l’une des causes de cette panne serait liée aux demandes des Entreprises européennes qui souhaitent que Google puisse garantir que les données restent hébergées en Europe.
Google prévoit, rapidement, de répondre à cette demande et travaille sur une partition de ses 40 centres de calculs mondiaux en grandes régions géographiques.

Google Apps Status Dashboard (Tableau de bord)

Pour permettre à ses clients de suivre en permanence les performances de toutes les applications de Google Apps, Google a ouvert à tous ses clients un accès à un tableau de bord qui donne, en permanence, une image de l’état de fonctionnement des applications.

Pour chaque incident répertorié, il est possible d’obtenir une explication détaillée et les prévisions concernant le retour à la normale.
J’imagine que ce tableau de bord était déjà disponible pour les responsables internes de Google ; l’ouvrir à leurs clients est une excellente idée.

La réponse de Google au SLA (Service Level Agreement)

Revevol étant l'un des clients de la version professionnelle de Google Apps, j'ai reçu un message d'excuses de Google expliquant les causes de la panne et me disant que ...
en allant bien au-delà de ses obligations, Google a décidé de prolonger gratuitement de 15 jours la durée du contrat de tous les clients.

En résumé : 2h30 de perturbations après plus de deux ans d'utilisation sans incident, 8 minutes effectives de désagrément, 15 jours de prolongation gratuite de mon service.
Combien d'informatiques internes, combien d'éditeurs traditionnels sont capables de proposer ce niveau de service ?

J’imagine déjà la réaction d’un directeur financier pragmatique : des pannes comme cela, j’en redemande !

Avec une panne tous les 15 jours, qui me perturbe pendant quelques minutes, j’aurai accès gratuitement à la solution professionnelle en permanence !!

Je crains, malheureusement pour lui, que ce directeur financier soit très déçu ; la fiabilité de Google Apps risque de l’obliger à attendre, longtemps, la prochaine prolongation de son contrat !

Gmail : la panne, expliquée et les clients, dédommagés ! - Blogs - ZDNet.fr

Gartner: Top 30 Countries for Offshore Service in 2008

Google Mots clés : Gartner: Top 30 Countries for Offshore Service in 2008. Offshore, Maroc Telecom, nearshore

Determining the country or countries that are best placed to host offshore IT operations is a daunting task for many organisations, according to Gartner, Inc. This year, Gartner has assessed the suitability of 72 countries as offshore locations, and has announced its ‘Top 30’. The analysis showed that the dynamic nature of the market has seen a number of countries position themselves as credible alternatives to the BRIC countries (Brazil, Russia, India and China).
“Countries such as Mexico, Poland and Vietnam have continued to strengthen their position against leading alternatives, while others have forced their way into the ‘Top 30’. These countries will be seeking to take advantage of the opportunity created by the increased focus that many organisations now have on cost optimisation, as a result of the current economic crisis,” said Ian Marriott, research vice president at Gartner.
During the last 12 months there has been significant activity in many countries to consolidate or grow their positions as leading locations for offshore services. “As a result of this, four countries have dropped out of the ‘Top 30’ and have been replaced by four that were just outside the ‘Top 30’ 12 months ago. This does not mean that the four ‘relegated’ countries have underperformed this year but the dynamic nature of the market has seen others making strong progress,” said Mr Marriott.
The four countries leaving the ‘Top 30’ this year were Northern Ireland, Sri Lanka, Turkey and Uruguay. The new entrants into the 30 leading countries for offshore services were Egypt, Morocco, Panama and Thailand. Strong interest in nearshore locations was a key factor; language skills, cultural compatibility, time zone and travel time were important considerations. As French speaking countries increase their proportion of work conducted offshore, they have been keen to find appropriate French language countries, and saw Morocco ‘step up’. The nearshore benefits of Egypt and Panama, and the cost consideration in Thailand were also important.    
In 2008, Gartner’s top 30 locations for offshore services, by region, were:
•    Americas: Argentina, Brazil, Canada, Chile, Costa Rica, Mexico and Panama
•    Asia/Pacific: Australia, China, India, Malaysia, New Zealand, Pakistan, the Philippines, Singapore, Thailand and Vietnam
•    Europe, the Middle East and Africa (EMEA): the Czech Republic, Egypt, Hungary, Ireland, Israel, Morocco, Poland, Romania, Russia, Slovakia, South Africa, Spain and Ukraine
Although only seven countries from the Americas appeared in the final list of 30, these countries are becoming an attractive proposition for the largest buying market for offshore services – the US. Only Canada was rated "excellent" for language (with fluent English and French) but Latin American countries are able to leverage their Spanish-language skills increasingly in the US as more organisations now require Spanish language from their providers for communication with parts of their workforce that speak Spanish as a first language.
The key evaluation criterion of cost was where Canada fared the worst ("fair") compared with "good" or "very good" ratings for all other countries in the region. However, Canada again led the rating for political and economic environment, cultural compatibility, global and legal maturity, and data and intellectual property security and privacy. Argentina was rated less favourably than the rest for its political and economic environment. Brazil and Mexico were considered "very good" for cultural compatibility, and the Latin American countries all managed a solid "good" rating for global and legal maturity. As observed in other regions, data and intellectual property security and privacy remain "work in progress" for many developing countries.
Ten countries from Asia/Pacific were represented in the 30 leading countries. These included the undisputed leader in offshore services — India — and the greatest challenger in terms of potential scale — China. The rest are a mix of mature environments that offer limited cost benefits (such as Australia, New Zealand and Singapore) and emerging countries with a variety of challenges, but attractive costs (such as Malaysia, Pakistan, the Philippines, Thailand, and Vietnam).
The final list of 30 countries included 13 from EMEA and for the first time saw two North African countries enter the leading countries in EMEA. Locations such as Ireland, Israel and South Africa fared well for language skills, because of the quality and quantity of English-language speakers. However, other countries, such as Morocco, Romania, the Czech Republic, Poland and Hungary were also given credit for the availability of alternative languages that address the needs of an increasing number of continental European buyers.
Cultural compatibility was variable, although only one EMEA country (the Ukraine) was rated lower than "good." In recent years, many countries in EMEA have become nearshore centres for traditional service providers and large Indian providers. This is reflected in the global and legal maturity section, where eight of the 13 countries scored between "good" and "excellent." Few countries in this region, with the exception of Russia, have a good selection of local service providers actively selling their capability outside their own country. In the final category of data and intellectual property security and privacy, a mature domestic environment or membership of the EU resulted in the highest ratings.
Gartner also found that external service providers (ESPs) have started to target places outside the ‘Top 30’ to get closer to mature countries, such as the Nordic regions and France that show increased interest in offshore. “Given the current financial turmoil, cost will remain an important factor. However having the right balance between lower cost and higher risks, and lower risks and higher costs will be critical in times of recession and uncertainty,” said Marriott.

Mobile banking in Africa

 

Fundamo was one of the first companies that demonstrated a working mobile payment solution with the first pilot in conjunction with Boland Bank announced during 2000. Absa Bank launched the first commercial mobile banking solution during 2002 in South Africa. Absa claims that more than 300 000 subscribers have enrolled for the service to date. This early launch have been followed up with launches of different types of solutions from Standard Bank, FNB, Nedcor, Investec and others. Although numbers are not readily available, a rough estimate of more than a million mobile banking subscribers is not far off. This is amazing if one considers the short time that mobile banking has been available.
During the past five years many initiatives were announced and often launched in South Africa. Many large technology companies and smaller ones made announcements and then disappeared again. South African companies that offered mobile banking solutions include Prism, Namitech, Cointel and Paym8. The launch of Wizzit in conjunction with Bank of Athens was so spectacular that they featured on CNN, Financial Mail and many other publications. Citibank and Standard and Chartered played with ideas to use the mobile channel to enter the South African market with a different and exciting angle, but did not implement.
The launch of MTN Banking during 2005 can be considered as the ultimate mobile banking solution with many new angles and an exciting offering. Combining mobile concepts with banking a totally new banking experience was created. Some of the concepts like "a banking starter pack", "pay as you bank" etc. worked well to grow a subscriber base quickly into hundreds of thousands of subscribers.
Today, South Africa is one of the leading examples of successful mobile banking deployments.
With the exception of South Africa and the successful deployments of Celpay in the DRC and Zambia (supported by Fundamo technology), quite a number of other examples of mobile payment and banking solutions exist in Africa. Celpay is one of the first mobile payment solutions deployed in the world and provides advanced payment functionality in countries with little (if any) payment infrastructure. It has been reported that Celpay payments in Zambia now accounts for almost 4% of the GNP.
Nigeria has turned into a hotbed of mobile payment innovation with one of the mobile operators Glo offering a product called Glomoney since launch. This solution provides mobile payment features on ATM cards from most banks. The solution is supported by one of the ATM switches called Interswitch who is now rolling this out to other banks and talking of providing it also via Celtel. (The solution initially available on special SIM cards, is now also available via a Java client to be downloaded on the phone). Other successful initiatives are from a company called eTranxact whom operates the solution themselves. The same technology is currently in production in Zimbabwe with Kingdom Bank. Flash-me-Cash is a SMS based viral payment mechanism that are claiming to have more than 500 000 subscribers and is based on a structured SMS solution.
Another area in Africa with a number of initiatives under way or in production is Kenya and surrounding countries. A number of banks recently announced mobile banking solutions. A noteworthy example is Consolidated Bank with a USSD mobile banking offering. South African company, Paym8 have deployed their solution since 2005 with Safaricom and is claiming an acceptable take-up. Card payment acquirer, iVeri (a Blue Label company) offers mobile payment functionality in Rwanda. One of the most important initiatives, however is the Vodafone backed mPesa initiative. This technology (developed in London) has been in pilot deployment since early 2006 and was recently placed in production with Vodafone subsidiary Safaricom. mPesa is receiving massive backing from London and is poised to be rolled out to other Vodafone networks (Egypt and possibly South Africa). The technology is SIM based and well designed with good “cash-out” functionality. A recently announced collaboration with Citibank allows for money remittance from London to Kenya.
A North African example is the roll-out of a mobile payment solution in conjunction with VISA in Morocco. The technology utilised was provided by French company Upaid and supported in country by Maroc Telecom and BCP (a bank). Other examples in North Africa are initiatives launched in Egypt and Tunisia.

A perspective on the History

The first mobile banking and payment initiatives was announced during 1999 (the same year that Fundamo deployed their first prototype). The first major deployment was made by a company called Paybox (largely supported financially by Deutsche Bank). The company was founded by two young German’s (Mathias Entemann and Eckart Ortwein) and successfully deployed the solution in Germany, Austria, Sweden, Spain and the UK. At about 2003 more than a million people were registered on Paybox and the company were rated by Gartner as the leader in the field. Unfortunately Deutsche Bank withdraw their financial support and the company had to reorganise quickly. All but the operations in Austria closed down.
Another early starter and also identified as a leader in the field was a Spanish initiative (backed by BBVA and Telephonica), called Mobi Pago. The name was later changed to Mobi Pay and all banks and mobile operators in Spain were invited to join. The product was launched in 2003 and many retailers were acquired to accept the special USSD payment confirmation. Because of the complex shareholding and the constant political challenges of the different owners, the product never fulfilled the promise that it had. With no marketing support and no compelling reason for adoption, this initiative is floundering at the moment.
Many other large players announced initiatives and ran pilots with big fanfare, but never showed traction and all initiatives were ultimately discontinued. Some of the early examples are the famous vending machines at the Helsinki airport supported by a system from Nokia. Siemens made announcements in conjunction with listed and high-flying German e-commerce company, Brokat. Brokat also won the lucrative Vodafone contract in 2002, but crashed soon afterwards when it run out of funds.
Israel (as can be expected) produced a large number of mobile payment start-ups. Of the many, only one survived – Trivnet. Others like Adamtech (with a technically sound solution called Cellpay) and Paytt disappeared after a number of pilots but without any successful production deployments.
Initiatives in Norway, Sweden and France never got traction. France Telecom launched an ambitious product based on a special mobile phone with an integrated card reader. The solution worked well, but never became popular because of the unattractive, special phone that participants needed in order to perform these payments.
Since 2004, mobile banking and payment industry has come of age. Successful deployments with positive business cases and big strategic impact have been seen recently.

Google Mots clés : Mobile Africa, MTN, banking, BCP, VISA, Maroc Telecom

Research confirming growth in Mobile Banking

Google Mots clés : Mobile, banking, Morocco

Recent research released in South Africa by World Wide Worx, have some very interesting findings about Mobile Banking in South Africa. The advantage of this research is that it has been conducted for the past three years. Clear trends can now be formed comparing the findings of 2005, 2006 and 2007. The results show an explosion in the usage by end-consumers of mobile banking. In this survey the number of respondents that have used mobile banking during 2007 have jumped to 17% (from 8%). In addition the research found that more than 50% of respondents plan to be using mobile banking by 2008.
However, the most interesting result of the research (which also tracks other services delivered via mobile phones) is that mobile banking is by far the mobile service used by older people. According to the research the likelihood of using cellphones for banking services increases with age, in contrast to the usage of other cellphone functions going down as users get older. For instance, urban cellphone users aged from 46 to 55 years are twice as likely to use cellphones for banking as those in the 19 to 24 age group.

La cata ! Gmail de Google est en

 

Google Mots clés : Gmail, la panne, Google SLA

par Louis Naugès:  panne ! Mardi 24 février 2009, 11h30, heure de Paris : Le monde entier s’inquiète, se désespère ! Un grand nombre des plus de 100 millions d’utilisateurs de Gmail, la messagerie Web gratuite de Google, n’ont plus accès à leur messagerie favorite ! J’en fais partie.

Pire, insupportable, inadmissible, cette panne touche aussi les utilisateurs professionnels de Google Apps, la solution payante (40 € /an / personne),  qui garantit un SLA (Service Level Agreement), en clair un contrat de garantie de service de 99,9 %.
Revevol, la société que je préside, est depuis quelques semaines, le premier distributeur européen de Google Apps ; nous avons aussi touché par cette “catastrophe”.

C’est le mal “absolu”, comme le titre avec beaucoup d’humour le site Web du journal Libération.

Les réactions du Web

Les statistiques de Twitter sont intéressantes ; pendant cette courte période de temps, le nombre de “twits” qui citent Gmail a explosé.

Le blog Techcrunch, que j’ai souvent cité, publie un texte en faisant, lui aussi, un jeu de mots, pour parler de “Gfail” (Géchec).
Beaucoup plus que le texte, ce qui m’a frappé, c’est le nombre de commentaires, qui dépasse déjà le chiffre de 170 !

Ce feu d’artifice de commentaires, blogs et autres twits c’est arrêté aussi vite qu’il avait commencé.
Maintenant que l’effervescence c’est calmée, quelles leçons peut-on tirer de cette panne de Gmail pendant 2 à 3 heures ?

Une analyse, à froid, des impacts de cette panne

Je vais concentrer mon analyse sur la version professionnelle, payante de Google Apps, qui comprend la messagerie.
En termes d’usages, quels ont été, pour moi, et toutes les personnes impactées, les impacts réels de cette panne ?

J’ai pu continuer à utiliser en mode normal, sans incident, toutes les autres fonctionnalités de Google Apps :
- L’agenda partagé.
- Les documents, traitement de texte, tableur, présentation.
- Google Sites, qui gère tous les Wikis de Revevol.
- Le partage de vidéo.

La seule fonctionnalité qui posait problème était donc la messagerie, dans sa version Webmail, accessible directement depuis un navigateur.
Je ne sais pas si c’est une coïncidence, mais les améliorations apportées récemment à Google Apps, et que j’avais présentées il y a quelques jours, m’ont permis de réduire très fortement les nuisances liées à cette panne.

Une analyse détaillée de la réalité des dysfonctionnements s’impose :

- Depuis mon Macintosh, j’ai mis en œuvre le mode de fonctionnement “flaky connection” (voir texte cité pour explication). Ceci me permettait de continuer à travailler avec ma messagerie professionnelle en mode off-line, mais en pouvant écrire de nouveaux messages ou répondre à mes courriels en attente.

Pendant toute la période de la panne, Olivier Margerand, consultant de Revevol, a mesuré en permanence la durée maximale de la période de non-connexion ; elle n’a jamais dépassé 8 minutes.
A chaque rétablissement, temporaire, de la connexion, j’ai reçu de nouveaux courriels et ceux que j’avais écrits ont été envoyés.

- Sur mon iPhone, en mode mobile, ma messagerie gmail a fonctionné en permanence.

- Les utilisateurs en mode iMap ou Pop3, dont je ne fais pas partie, m’ont aussi confirmé qu’ils ont pu continuer à utiliser leur messagerie gmail professionnelle.

En résumé, pendant 3 heures, l’impact négatif réel le plus grave de cette panne, sur mes usages de Google Apps, c’est résumé à :

Un délai maximum de huit minutes pour l’envoi ou la réception de mes messages.

Comment résumer cette journée “cata” !
- J’ai été gêné par une panne de Gmail : oui.
- Ma capacité à envoyer et recevoir des courriels a été sérieusement impactée : non.
- J’ai pu continuer à utiliser toutes mes autres fonctions “SaaS” sur le Cloud de Google sans aucune difficulté : oui.

Ce ne sont pas, objectivement, des perturbations “catastrophiques” !
Le bruit fait par cette panne dans la “blogosphère” est ... disproportionné par rapport à la réalité des perturbations ; comment l’expliquer ?

J’y vois plusieurs raisons :
- Plus rien de ce qui touche à Google ne laisse indifférent.
- Tout incident de ce type ravive les craintes liées à la migration des applications dans le “Cloud”.
- Les pannes SaaS sont si peu fréquentes qu’elles font la “une” à chaque fois.  Il y a deux mois, c’était Salesforce qui était montré du doigt pour un arrêt de service inférieur à 1 heure.

Les réponses de Google

Sur le blog officiel Gmail, Google a publié deux textes, le premier pour annoncer la panne, le second pour indiquer que tout était rentré dans l’ordre.

Google a aussi, et c’est normal, transmis ses excuses à toutes les personnes qui ont souffert de cette panne.

D’où est venue cette panne ?

J’avoue avoir du mal à en comprendre la nature.
- Ce ne sont pas les serveurs, car toutes les autres fonctions de Google Apps étaient opérationnelles.
- Ce n’est pas l’application Gmail elle-même, car elle continuait à fonctionner sur iPhone et en mode iMap.

Google, dans sa réponse, indique que, pour le moment, ils n’ont pas encore trouvé la cause de cette panne ; j’attends avec impatience les explications.

Et si Google, dont tout le monde connait la devise “Don’t do evil” l’avait fait exprès pour montrer l’efficacité de ses nouvelles options off-line et sur mobile.
J’avoue que cette pensée m’a, quelques secondes, quelques secondes seulement, effleuré l’esprit !