Archive for the 'Amazon EC2' Category

Consoles, Management Tools and Brokers

September 15, 2009
Written by George Reese

The web interfaces of cloud providers such as Amazon Web Services (AWS) have consistently lagged behind third-party consoles in both user experience and functionality. Third party consoles supported rich meta-data around AWS resources in addition to  integrating CloudWatch and CloudFront support long before the AWS console.

The initial lack of capabilities in the provider interfaces created the market for cloud consoles. The initial crop of cloud consoles, however, has grown beyond simply providing a better interface into the cloud. This market has evolved into cloud brokers that act as the trust layer between the customer and their cloud infrastructures. The evolution from cloud console to cloud broker occurred in less than a year. In 2008, the only way to access Amazon Web Services was a command-line tool or writing your own code. As a result, a number of people built their own custom web interfaces for accessing AWS and a number of third-party tools like enStratus appeared in the market to simplify the job.

As the market changed, the role of a pure console became obsolete. Amazon added a graphical console for AWS. Today, all cloud providers have some form of basic web interface into their services. Though most third-parties still provide better front-ends with more features than the provider consoles, the third-parties have become something much more significant: cloud management tools and cloud brokers.

User Manager Console

A cloud management tool adds functionality beyond the simple management of raw cloud assets. RightScale, for example, provides robust auto-scaling beyond what AWS auto-scaling allows. enStratus, on the other hand, provides key management that enables complete encryption of data at rest and in transit in the public cloud without the cloud provider being able to gain access to your keys.

Cloud Server Console

Cloud brokers offer yet another layer of functionality on top of cloud management tools. A cloud broker provides a single interface through which you can manage multiple clouds and share resources across clouds. A cloud broker provides the following capabilities:

  • It provides a single interface for interacting with multiple clouds
  • It operates outside of the clouds it controls and monitors those clouds
  • It detects cloud failures and reacts in some appropriate way to those failures
  • It can move infrastructure elements from one cloud (public or private) to another

A cloud broker is a critical piece of your public cloud infrastructure. Minimally, a cloud broker provides a solid business continuity strategy. Without a cloud broker, your infrastructure is at risk against failures in your cloud provider.

The ultimate failure is, of course, the cloud provider going out of business. Outside the cloud, you are somewhat isolated from vendors going out of business. For example, if your non-cloud infrastructure is built entirely on Windows, your Windows desktops and servers will still start up if Microsoft were to declare bankruptcy. If, on the other hand, your infrastructure is running entirely in Azure and you have no cloud broker, you go out of business along with Microsoft. If you are a public company, the argument that “Microsoft will never go out of business” is not sufficient for SOX compliance. Even if it were (or if you are a private company), you should keep in mind that no one ever thought Enron would go out of business or that GM would declare bankruptcy.

A cloud broker can spread your operations across multiple cloud infrastructures and enable real-time disaster recovery into a secondary cloud. If Amazon were to suddenly disappear for some arbitrary reason, enStratus can enable you to bring that infrastructure back up in Rackspace; for example, with little (or even no) data loss.

In the future, cloud brokers should evolve into brokers in the truest sense of the word “broker”. The Holy Grail in the cloud broker world is the ability to automatically decide the least expensive cloud components for efficiently operating your systems and move those components around different public clouds based on pricing and quality of service. Though we’re still a long way from that vision, cloud brokers are creating the building blocks that will make the vision possible.

George ReeseGeorge Reese is the CTO of enStratus™, the provider of the leading cloud management platform for enterprise applications. George is also the author of the #1 selling cloud computing book “Cloud Applications Architecture“. To learn more about enStratus visit www.enstratus.com


Deploying a Web Server on Amazon EC2

April 30, 2009
Written by Ameed Taylor

Detailed tutorial on setting up and deploying a web server on Amazon EC2.


Amazon Elastic MapReduce Beta Launched

April 2, 2009
Written by Ameed Taylor

elasticmap

Amazon Web Services today announced the public beta of its newest addition to its web services portfolio; Amazon Elastic MapReduce.

Elastic MapReduce is a web service that allows researchers, analysts and businesses to process and analyze data using a hosted Apache Hadoop “data processing engine” that runs on Amazon S3 and Amazon EC2.

Much like the Mathematica and Mathworks EC2 computational solutions, Amazon Elastic MapReduce is a metered service that is designed to be used for applications like data mining, machine learning, financial analysis or scientific research. Amazon’s entry into the market will provide a bit of competition (to say the least) to Cloudera for users interested in running Hadoop in the Enterprise.

Elastic MapReduce will initially only be available in the USA and users can choose standard Amazon EC2 Instances or High CPU EC2 Instances to run their applications. Pricing for Elastic MapReduce ranges from 10 cents per hour for a small standard EC2 instance to 80 cents per hour for an extra large High CPU instance. Elastic MapReduce supports Ruby, Python, PHP, Java, and Perl.

Elastic MapReduce will be appealing to users that need to scale up quickly as the High CPU plans allow users to run up to hundreds of instances running Hadoop and instances can be run concurrently. Amazon EC2 and S3 have to date been very stable and the fact that EC2 and S3 are integrated with Elastic MapReduce and that instances can be managed via the AWS Console or command line are key selling points.


Discount Passes for Cloud Slam09 Conference

March 31, 2009
Written by Ameed Taylor

CloudSlam09

Cloud Slam’09 is a virtual cloud computing conference taking place over 5 days from 20 – 24 April 2009.

The conference has a number of tutorials, workshops and panels broken down into 5 tracks. (Cloud Technology, Cloud Implementation Experiences, Cloud Privacy and Compliance, Cloud Business Models and Cloud Computing Research)

Aimed at C level executives, developers and cloud end users, Cloud Slam’09 has a number of high profile speakers scheduled to take part including Simon Crosby, Dr.Werner Vogels, Russ Daniels and Stephen Herrod.

OnDemand Beat readers can receive 5% off the conference registration fee by using discount code ONDEMANDBEAT50