Search results for 'app:weblogs' matching tag 'Cloud Computing'

La caída de gmail, lecciones aprendidas

lopez — Wed, 02 Sep 2009 05:00:00 GMT

Ayer, martes 1 de septiembre, se cayó el servicio web de gmail, por un tiempo (no tengo el dato exacto de falla, pero fueron por lo menos dos horas). Yo uso el servicio via web, y tengo varias cuentas, alguna personal y otras por empresa cliente. Y me parece que el servicio de gmail es muy bueno, y la interface de manejo web me resulta útil: el tener “labels” por ejemplo, para manejar los temas, es algo que deberían tener todos los programas de correo desde hace años: basta eso de que un email va en una carpeta. Es como el delicious: cada item interesante, puede clasificar en más de un tag (es idea a explorar, que tengo pendiente desde hace unos años, desde que armé mi sitio, vi delicious, y estudié topic maps y temas relacionados).

Es interesante leer en el blog de gmail, las causas y medidas que se tomaron para subsanar la falla en el servicio, por ejemplo, leo en:

http://gmailblog.blogspot.com/2009/09/more-on-todays-gmail-issue.html

Here's what happened: This morning (Pacific Time) we took a small fraction of Gmail's servers offline to perform routine upgrades. This isn't in itself a problem — we do this all the time, and Gmail's web interface runs in many locations and just sends traffic to other locations when one is offline

Bien, la “redundancia es la mejor estrategia”, una frase de “Sadrac en el horno” de Silverberg, es una frase que parece que Google sigue siempre.

However, as we now know, we had slightly underestimated the load which some recent changes (ironically, some designed to improve service availability) placed on the request routers — servers which direct web queries to the appropriate Gmail server for response. At about 12:30 pm Pacific a few of the request routers became overloaded and in effect told the rest of the system "stop sending us traffic, we're too slow!". This transferred the load onto the remaining request routers, causing a few more of them to also become overloaded, and within minutes nearly all of the request routers were overloaded. As a result, people couldn't access Gmail via the web interface because their requests couldn't be routed to a Gmail server. IMAP/POP access and mail processing continued to work normally because these requests don't use the same routers.

Vean cómo la liebre salta cuando menos se lo espera. “Esperar lo inesperado” sería algo que resume la lección aprendida. Si tenemos un sistema andando, sea software, hardware, y demás, habrá que estar preparados para cuando pase lo que no debería pasar.

The Gmail engineering team was alerted to the failures within seconds (we take monitoring very seriously). After establishing that the core problem was insufficient available capacity, the team brought a LOT of additional request routers online (flexible capacity is one of the advantages of Google's architecture), distributed the traffic across the request routers, and the Gmail web interface came back online.

Ciertamente, cuánto de nosotros planeamos las aplicaciones para que sean monitoreables? ¿o tenemos preparados los recursos para que sea tan flexible la arquitectura que permite un reemplazo rápido de los mismos? Yo debería aprender más sobre las técnicas de monitoreo, protocoles, estándares, que existen en la industria.

Vean igual que el problema se circunscribió a la interfaz web: el acceso IMAP/POP siguó funcionando.

Algunos usuarios en Twitter, empezaron a desgañitar pestes sobre google y gmail. Pero me parece la gente de Google hizo un excelente trabajo. Tengo menos caídas en Gmail, que en servidores locales de empresas cliente, que atienden a apenas decenas de usuarios.

Pienso entonces: si bien esto se puede ver como una señal de fragilidad de un servicio en línea (y de lo que pronto va a ser llamado algo difusamente “la nube”), veo que es una señal más de lo bueno que es delegar la solución de problemas a gente que está preparada para eso. ¿cuántos de nuestros servicios IT que consumimos en empresas a las que vamos todos los días, podrían reaccionar tan rápido? Yo he visto sistemas miles de veces más chicos, caídos por horas (o días), acá en estos lares.

Una opinión en contrario:

http://www.nevillehobson.com/2009/09/02/the-cloud-is-still-not-reliable-enough/

donde leo:

I still have more faith in local content and offline backups.

¿Opiniones?

Nos leemos!

Angel “Java” Lopez
http://www.ajlopez.com
http://twitter.com/ajlopez

Qué es Cloud Computing

lopez — Fri, 21 Aug 2009 05:00:00 GMT

Hay bastante revuelo sobre los conceptos y aplicaciones de Cloud Computing, con ofertas de prácticamente todos los grandes jugadores del software. En estos días encontré este video, corto, que sirve para que podamos transmitir, no tanto el tema técnico, sino la proposición de valor que ofrece la nube. Podemos verlo como una extensión de tercerizaciones que ya se venían haciendo (como el hosting del sitio de una compañía), pero a mayor escala (a tener en cuenta para evaluar este video: es un video comercial de SalesForce.com):

(Please visit the site to view this media)

Hay muchos temas técnicos a discutir, y muy interesantes. Por ahora, vayamos viendo cómo puede influir en nuestros desarrollos y proyectos, todo esto de la nube.

Nos leemos!

Angel “Java” Lopez
http://www.ajlopez.com
http://twitter.com/ajlopez

Windows Azure SDK July update

NunoGodinho — Wed, 29 Jul 2009 05:00:00 GMT

Windows Azure has a new SDK release that is now available. A very interesting new feature is the possibility of having several Roles per deployment, unlike what happened in earlier releases where we would only be able to have one Web Role and also a one Worker Role. Another interesting feature is now we have the possibility to chose between a normal ASP.NET Web Forms application and a ASP.NET MVC application as the Web Role.

Download the Visual Studio Tools here.

Download the Windows Azure SDK here.

More on Windows Azure Commercial Availability

NunoGodinho — Wed, 22 Jul 2009 05:00:00 GMT

Microsoft announced on 14th July the commercial availability of Windows Azure as I posted previously and now there are more information’s about the the same subject so I decided to write again on the subject.

Important elements to retain:

“During Community Technology Preview (CTP), services included in Windows Azure will be available without charge - subject to certain limits:

Total compute usage: 2000 VM hours

Cloud storage capacity: 50GB

Total storage bandwidth: 20GB/day“

How Consumption is Measured:

How Consumption is Measured?
- Windows Azure
  - SLA
    - “(…)we guarantee that when you deploy two or more role instances in different fault and upgrade domains your Internet facing roles will have external connectivity at least 99.95% of the time (…)”
  - Compute time, measured in machine hours
  - Storage, measured in GB
  - Bandwidth requirements (transmissions to and from the Azure datacenter), measured in GB
  - Transactions
- SQL Data Services
  - SLA
    - “(…) will maintain a “Monthly Availability” of 99.9% during a calendar month. “Monthly Availability Percentage” for a specific customer database is the ratio of the time the database was available to customer to the total time in a month. Time is measured in 5-minute intervals in a 30-day monthly cycle. (…)”
  - Web Edition Relational Database includes:
    - Up to 1 GB of T-SQL based relational database
    - Self-managed DB, auto high availability and backup
    - Auto Scale with pay-as-you grow
    - Best suited for Web application, Departmental custom apps.
  - Business Edition DB includes:
    - Up to 10 GB of T-SQL based relational database
    - Self-managed DB, auto high availability and backup
    - Auto Scale, Pay-as- you grow
    - Additional features in the future like auto-partition, CLR, fanouts etc.
    - Best suited for ISVs packaged LOB apps, Department custom apps
- NET Services
  - “(…)Uptime percentage commitments and SLA credits for .NET Services are equivalent to those specified above in the Windows Azure SLA. Due to inherent differences between the technologies, underlying SLA definitions and terms differ for .NET Services. Using the Service Bus module of .NET Services, customers will have connectivity between a customer’s service endpoint and our Internet gateway(…)”
  - Messages are charged to the customer in discrete blocks of 100,000 (“100k”) for each monthly billing period, meaning that
  - A customer who consumed 95,000 messages would be billed for 1x100k messages (plus the bandwidth used to send messages in or out).
  - A customer who uses 150,000 messages in a billing period would be charged for 2x100k messages (plus the bandwidth used to send messages in or out).
  - A customer who uses 20 million messages in a billing period would be charged

More information can be be found on the Windows Azure site.

Windows Azure Commercial Availability and Business Model announced

NunoGodinho — Tue, 14 Jul 2009 05:00:00 GMT

Just today in the WPC 2009 Microsoft announced the Commercial Availability and Business Model that will be used in Windows Azure and Windows Azure Service Platform, and so a brief info about the amounts involved.

Windows Azure
- Compute – 0.12 $ per hour
- Storage – 0.15 $ per GB by month stored
- Storage Transactions – 0.01 $ per 10K

SQL Data Services
- Web Edition – up to 1 GB relational Database – 9.99 $ per month
- Business Edition – up to 10 GB relational Database – 99.99 $ per month

.NET Services (includes Service Bus, Access Control)
- Messages – 0.15 $ per 100K message operations

Bandwidth across all three services will be charged:
- 0.10$ for inbound communications
- 0.15$ for outbound communications

Guarantee Availability:

at least 99,95 % for external connectivity
at least 99,9% for storage

Availability at launch will be Australia, Austria, Belgium, Canada, Denmark, Finland, France, Germany, Ireland, India, Italy, Japan, Netherlands, New Zealand, Norway, Portugal, Spain, Sweden, Switzerland, UK, and the United States.

In the March 2010 timeframe we expect commercial availability to expand to Brazil, Chile, Colombia, Czech Republic, Greece, Hong Kong, Hungary, Israel, South Korea, Malaysia, Mexico, Poland, Puerto Rico, Romania, Singapore, and Taiwan.

More information's here.

Windows Azure Services Platform Changes

NunoGodinho — Wed, 08 Jul 2009 05:00:00 GMT

.NET Services July 2009 CTP Release

An one more CTP of Windows Azure was made available and this one has some interesting points like Windows 7 RC Support, easier setup, but of course the large amount of changes is in terms of connecting heterogeneous environments, in terms of:

Connect Applications and users
Interoperability by Default, very important since it supports now even more standards than before
Federation of Data, Messages, Identity and Access, that makes solutions more easier to interoperate between each other.

Of course this changes occur in the spectrum of the .NET Services Bus and Access Control, since the Workflow Services is now offline in order to incorporate more customer feedback.

Check more information here.

SQL Data Services

Now we can look at SQL Data Services as a real Database, and now give support for TSQL in SQL Data Services, of course in this release we only have a subset of TSQL, but this is a good way to go.

Of course there are differences between SQL Server on-Premise and SQL Data Services and some of they are:

“(…)

SDS is a multi-tenant system

The hardware resources are owned, hosted and maintained by Microsoft

SDS is a service

(…)”

So what’s in?

DDL
- FUNCTION
- INDEX
- PROCEDURE
- ROLE
- SCHEMA
- STATISTICS
- SYNONIM
- TABLE
- TRIGGER
- VIEW

DML
- SELECT/INSERT/UPDATE/DELETE functionality
- DML triggers
- JOINs
- Transactions
- A large number of built-in functions (aggregates, math, date and time, ranking, etc.)

Manageability
- Query tuning via SET SHOWPLAN and SET STATISTICS
- Index tuning via create and drop index
- Statistic management via UPDATE STATISTICS
- Information schema views and system catalog views (e.g., sys.databases)
- Local HA is automatically provided as part of the service.
- Upcoming version of SSMS and VS tools will work against SDS

Ok very nice v1 of SQL Data Services, check more in this post from the SQL Data Services Team.

Notes from the Windows Azure by David Chappell in ArCast.TV

NunoGodinho — Tue, 07 Jul 2009 05:00:00 GMT

I was viewing this ArCast.TV chapter about Windows Azure by David Chappell and I found that it would be interesting for me and for other people to get some notes about this, and so this post was created due to it.

We see in Cloud Computing two types of elements:

SaaS – Software as a Service

Examples:

SalesForce.Com
Microsoft

CRM Online (competitor of SalesForce.com)
Exchange Online
Sharepoint Online
and so on.

PaaS - Platform as a Service or Cloud Platforms

Competition in this space

Amazon (1st company in this space)

EC2 – we get Virtual Machines

Provide Virtual Machines in the Cloud that can be Windows or Linux, and after that those are your VM’s that you have to manage. Very different approach from Windows Azure.

Microsoft

Windows Azure - We give the Applications and the Virtual Machines are managed for us, without we ever need to manage a specific Virtual Machine

Components:

Windows Azure – Think of it as a Windows in the Cloud, of course this is not the same thing but is a nice way to make the analogy to it
Windows Azure Service Platform

.NET Services

Service Bus
Workflow Services
Access Control

SQL Data Services
Live Services

Google

AppEngine – Not like EC2 but more like Windows Azure

We give a Java or Phyton application and the will be place on VMs like the Windows Azure, but this is completely transparent for user
All about Supporting Web Applications
Not possible to have a Batch Service like in Azure
Only 10 seconds available to process Responses, or else the AppEngine will shutdown the Request

What’s the best? EC2, Windows Azure or AppEngine?

It depends on the application your building and the way you’ll build it because the approaches are different and so it’s important to get the most correct Technology to your problem.

ISV’s Strategy

In case of a Windows ISV then is normal to think about Windows Azure in order to address the SaaS space.

Enterprise Strategy

Often enterprises look at the cloud only as a way to save money, and move the current apps unchanged into the cloud, but that is not the purpose of Windows Azure since now this is not possible, but of course there will be attractive to move existing applications into the cloud.
Of course all depends on the applications, because it’s possible if we have a Web Application with a SQL Server Database, because in this case we can move the Web Application to the Windows Azure and the SQL Server Database into the SQL Data Services.

What this makes?

Application more manageable
More Scalable
More Extensible
and so on.

Get Started using Windows Azure in www.azure.com.

Windows Azure Service Platform: July CTP Breaking Changes Announcement

NunoGodinho — Mon, 06 Jul 2009 05:00:00 GMT

The Windows Azure Team has just release a July CTP Breaking changes announcement since the already breaking changes in the Workflow Services, now there are changes also in the .NET Service Bus.

“(…)Queues and Routers data will NOT be persisted and restored after the maintenance. Users will need to back up their data if they wish to restore them after the July 2009 CTP release. Please see below for detail.

As previously announced, the existing Workflow Service will be removed from .NET Services in the July 2009 CTP release. Any solutions that currently rely on the Workflow Services will have to be modified on or before 7/7/2009 9am PST in order to continue functioning smoothly. Existing solution Workflow Service metadata such as Workflow Type will also be deleted and cannot be retrieved after the July 2009 CTP release. (…)”

Impacts on:

“(…)NET Services and the .NET Services Portal will be unavailable during this period.(…)

Experiments with the new SQL Data Services

NunoGodinho — Mon, 06 Jul 2009 05:00:00 GMT

The SQL Data Services that are part of the Windows Azure Service Platform just recently, and due to customer feedback, changed the way of working from a ACE (Authority, Container, Entity) model to a TDS (Tabular Data Stream) model.

Eugenio Pace in this post show how he changes the IssueTracker application from the old model to the new one, and also showing the things missing from one to the other and what needs to be done.

Very interesting reading indeed.

Windows Azure Service Platform – Worflow Services Changes

NunoGodinho — Mon, 06 Jul 2009 05:00:00 GMT

The team announced that several importance changes in the Workflow Services way of work and also that it will be taken offline for a while in order to proceed with all the changes.

Changes like the possibility of working in the Cloud Workflow Services the same way as in the .NET Framework 4.0:

“(…) An area of consistent discussion is the Microsoft .NET Workflow Service delivered via .NET Services, and how it relates to the Windows Workflow Foundation (WF) in the .NET Framework. One of the comments that we’ve consistently heard about the .NET Workflow Service is that you want the Workflow Service to be built on .NET Framework 4‘s workflow engine. This is currently not the case, since we are prior to the release date of .NET Framework 4.

As the direct result of user feedback, we will hold off further releases of the Workflow Service until after .NET Framework 4 ships. Since there will be important changes to the Workflow Service before it goes to full production, we are planning to take down the existing Workflow Service as part of service improvements in the month of July. This means any solutions that currently rely on the Workflow Service will have to be modified on or before July 1 in order to continue functioning smoothly. (…)

More information’s here.