Hype and high expectations have always been there as an important risk for the Cloud to never overcome the high expectations created. We have got used to impressive IaaS system, delivering easily-manageable virtual infrastructures, virtually infinite resources, etc. or advanced PaaS Clouds letting us to deploy our applications and easing data persistence and other development-related tasks. However, the end user was often overlooked and the Cloud had few examples at a global scale aimed at satisfying end users’ needs.
Today, MORFEO Cloud technologies is proud to announce the result of some years of research and tests in which members of this chapter actively collaborated.
We’d like to present 3GBox, also in the Mobile World Congress 2010. Being more than a 3g modem, 3gBox helps users to store data in the cloud; their SIM card becomes the key element for security and a local cache is in charge of storing data to upload the information to the Cloud depending on the available bit rate. Also, heuristics are implemented that help keep the most “useful” data locally stored to improve users’ experience. Updates,O.S. drivers and so on are downloaded from the Cloud to help usage and configuration.
Yes, I was trying to attract your attention with a “British tabloid” style heading here. The news is not at all so radically new, but now that I attracted your attention I beg you pardon and kindly ask you to allow me 4 minutes to sum up Amazon’s “new” billing model 
Amazon announced spot pricing for cloud compute instances. EC2 customers can indicate their own price, and Amazon EC2 will bring compute instances up at variable discount prices according to these “bids” [1].
This move is in sync with their strategy extra-cost reserved instances, which is regarded as an evolution by many, but, frankly resembles previous allocation models in Grid computing. Again, nothing new under the sun. Indeed, auction systems supported by software agents and expert systems have been in the market for long long time.
In [1], the authors raise a very interesting question that we generalize and rephrase here: Are different billing models needed for different Cloud service types? How many do we need per service type? Is the billing model the only important parameter here?
From this humble Internet corner, we bet that automated bidding systems will play a role for massive service provision and better prices to be acquired. Still, some important features are still missing such as for instance custom billing support for “VIP” clients.
[1] http://web2.sys-con.com/node/1220487
A recent article describes “new” approaches by Amazon, CohesiveFT and other relevant Cloud players to secure their networks.After all, it is not just about providing secured VMs in a very secure hypervisor, but also about securing the very communications themselves.
Again, the wheel turns to use well-known technologies for securing the networks and label them with the new buzzword.
In addition, Amazon offers the possibility of integrating a corporate physical machine with those in the public Cloud so as to let users keep the sensitive data home. A deeper analysis on how the information exchange is done between local physocal machines and Cloud machines, how keys are maaged and so on.
Anyway, this new announcement shows the important concern security has become when it comes to Cloud adoption by enterprises.
Very recently, a well-known researcher in the Grid community, Thomas Sandholm, published a paper in the First International Conference on Cloud Computing, held in Vancouver.
The scientific contribution of the paper is very limited and it does not mean to be a full state of the art on Cloud technologies, as recognized by the very same authors. However, this work draws a nice architectural design and maps currently existent products to the different layers of the architecture.
Importantly, the authors identify a gap that fits well in the interest of Telecom operators. Administration and Business support tools still lack an appropriate offer to meet enterprise requirements. Deployment, configuration, monitoring and life-cycle managent at the SaaS level need support from the underlying PaaS layer. The PaaS (e.g. programming and execution environments) and high infrastructure services (e.g. Google Bigtable or Amazon’s Dynamo) layers of a complete Cloud architecture are key enablers for an appropriate scalability of the system: Cloud scaling power is not just about placing new (virtual) hardware and deploying services in there. Scalability also needs to be supported by really scalable data or event PaaS/High Infrastructure-level services.
Even though this may seem way too obvious, we have recently seen many examples of Cloud architectures overlooking or disregarding the creation of “scalability-enabling” services. We strongly believe these elements are the cornerstone to offer actually scalable services to millions of end-users over our Cloud.
A recent study by Gartner suggested that software vendors are not satisfied with the SaaS approach since it “is not quite the panacea it often promised to be”. Developers have realized that to deliver services on the Cloud they have to make the same huge efforts towards, for instance, scalability, security and data management they did before the Cloud. Is offering appropriate PaaS means underlying the SaaS layer the actual panacea?
A recent MIT Technology Review reveals how IBM’s strategy to compete with the likes of Amazon.com and Salesforce.com is to employ a tailored Cloud, designed to work for specific types of tasks, rather than offering generic storage and processing that can be used for whatever a customer need.
Read the rest of this entry »
