Three months ago, we published an article in the prestigious ACM’s Computer Communication reviews that dealt with the concept of Cloud Computing. The importance of this article has been acknowledged by the community with a handful of references, citations and downloads.

The most striking figure is that of software downloads from the ACM Digital Library.  More than 1200 downloads in three months is certainly a honor for us.

Although it is quite difficult to get references in such short time periods (just three months), the article has already been referenced in a recent article in the IEEE Computer Society Technical Committe on Data Engeneering.

The pedagogical insight has been demonstrated by the inclusion of the article in the lectures at many universities and important sources worldwide, for instance:

• Zürich
• St Andrews
• Wikipedia
• Valladolid
• Montevideo

Extra Academical entities have also relied on our definition for their work, McKinsey report Clearing the air in Cloud computing  has emphasized the work done by Telefónica towards avoiding hype and finding a reasonable definition of the Cloud. Also, Peeyush Tugnawat’s blog from Oracle has also recognized the value of the article.

We would like to thank this wide support and acknowledge the work done by all the authors our work relied on, one can only build on other’s work. Thanks a lot!!

Luis Vaquero, Luis Rodero, Juan Cáceres

P.S. We’d also  like to thank Maik Linder (from SAP Research, co-author of the paper). Sorry guy, we did not forget about your nice work here

Current commercial IaaS (Infrastructure as a Service), like Amazon EC2, Felxiscale, GoGrid, etc., provide hosting clouds solutions based on computing, storage and networking virtualization. Through proprietary APIs and management tools, normally offered on Internet, customers can manage virtual machines images with an specific OS installed, and ask for their deployment in the Cloud. These cloud providers bill for the usage of infrastructure in a pay-per-use way.

But, are current Hosting Clouds enough for deploying Enterprise Services?

Read the rest of this entry »

In the general context of cloud interoperability (to decouple the application the cloud user wants to deploy from the cloud infrastructure to which that deployment is issued), OVF has been suggested several times as a very promising candidate. Using this standard (produced by the DMTF), virtual machine based services can be packaged in a vendor a platform neutral way, which makes it very suitable for IaaS clouds.

This “statements of interest” include the Citrix partnership with rPath to “extend Kensho [the flagship OVF-related public Citrix project] to support the deployment of OVF appliances in infrastructure clouds, starting with Amazon EC2”, the Oracle and Intel collaboration to “extend standards that enable portability of virtual machines images such as Open Virtual Format (OVF) and also create Web services standards for provisioning and management of cloud-based services” and the VMware’s vCloud initiative that “will build upon that work [OVF] by submitting a draft of its VMware vCloud API to enable consistent mobility, provisioning, management, and service assurance of applications running in internal and external clouds”.

Although support from the main industrial players is undoubtedly important, these statements basically are no more than declaration of intentions (with the possible exception of vCloud API, but which is too obscure yet to have a precise idea about it). They don’t go further analysing the (not trivial) problems of OVF applied to cloud infrastructure. It is worth mentioning that OVF was not designed with cloud computing in mind, so there are issues that need to be solved when applied to this environment, in particular, on automatic elasticity (OVF is designed for static deployments only), self-configuration (OVF is designed to use customization parameters that are known in pre-deployment time, which is not realistic in clouds) and deployment constraints (OVF don’t include any mechanism to provide hints for deployment, that are specially relevant in clouds due to business criteria, e.g. the database with the data of my customers can not be deployed outside my country due to regulations).

The RESERVOIR project (focused on federation of clouds at infrastructural level) is doing so. We have analysed the aforementioned problems and we are proposing extensions to OVF to solve them. This is not just theoretical work, because it is backed by an actual cloud middleware prototype. Eventually those extensions could be proposed to DMTF to enhance the OVF specification (now that even Winston Bumpus, the President of the DMTF itself, has expressed the OVF target in cloud, describing the standard as “an ideal cloud migration and deployment package” in the recent SATCCI Workshop) when the appropriated maturity level gets achieved.

By the moment, we have submitted a paper to COMSWARE’09 (authored by Telefónica, SAP, IBM and Sun, four major companies in the ICT industry) performing an early analysis of the problems in detail and proposing solutions (among all the possible ones) to adapt OVF to clouds. It has been accepted for publication, so by mid June the details will be disclosed and we will be very happy to share this knowledge and get feedback for others working on the interesting and hop topic of cloud interoperability.

New research from McKinsey & Co. prompts that adopting a cloud model would be a money-losing mistake for most large corporations. “Clearing the Air on Cloud Computing,” concludes that outsourcing a typical corporate data center to a cloud service would more than double the cost. Using Amazon.com’s as pricing reference, the total cost would be $366 a month per unit of computing output, compared with $150 a month for the conventional data center.

McKinsey claims that owning the hardware is cost-effective for most corporations when the depreciation write-offs for tax purposes are included. They also believe that the labor savings from moving to the cloud model has been greatly exaggerated.

We have to strongly argue about this type of utilization of the cloud by big corporations. The real magic of the cloud lies beneath its skin. A cloud is not just a conventional hosting environment. It provides advanced monitoring capabilities that allow you to automatically control how the service grows. In other words, clouds are best fitted for Dynamic environments.  In these, services’ load can hugely vary over a short period of time and the company may be aware of that beforehand, thus reserving capacity in advance to attend the expected peak or rapidly reacting to unexpected demand growths. Let us propose two examples for this.

1. Cell network degradation during new year’s eve. The Telco operator would rent the required infrastructure to operators worldwide (Australian load would be minimum at that time and prices and resources availability should be ok there), thus, facing the increase in a cheap and effective manner.
2. Special promos made by third parties may cause a sudden increase in services’ demand, with advanced cloud monitoring systems we can detect the increase and quickly react to accommodate as much demand as possible in order to maximize the benefit (and the quality of user experience) while keeping the infrastructure to a minimum (and is associated costs)

Summing up, while it is clear that a cloud is a very expensive 24/7 environment, we strongly believe (our preliminary data confirm this) that short term outsourcing should be made in a cloud environment in order to get the required QoS level and minimize the costs associated to sudden or expected short load increases.

We would like to welcome the Galicia Supercomputing Center (CESGA) to the MORFEO Cloud Technologies Chapter. CESGA is the HPC and advanced services Data Center for the Galician Scientific Community, University Academic System and the Spanish Scientific Research Council (CSIC).

CESGA has recently designed and integrated the Finis Terrae, one of the biggest Supercomputers in Spain (second in float numbers processing power) and one of the biggest European shared memory supercomputers.

In parallel to the communications and supercomputing services, CESGA is developing an intense R&D activity in Infrastructure Provisioning-related topics. CESGA has collaborated in projects such as EGEE, int.eu.grid, CEAPTB, TORGA.net, EvoProc and Formiga.

Carlos Fernández will represent CESGA at the MORFEO Cloud Technologies chapter.

Welcome again!!