Cloud Insights from Intel Developer Forum 2010
I was fortunate enough to attend Intel’s Developer Forum in mid-September at the Moscone Center in San Francisco to attend most of the cloud-related sessions. It was amazing to see the various innovations displayed by Intel IT with regards to cloud architectures and how they make business more agile. Amongst my findings:
- Intel IT had narrowed its service provisioning time from 90 days in 2009 to 14 days in mid-2010 and have a goal of 3 hours by end of 2010
- They are doubling their rate of virtualization and hope to have 70-80% of their servers virtualized by 2011
- A ‘Hosting Automation Framework’ allows users (consisting of three audiences – App Developers, IT Customers, and IT Operations) to access services through a self-service portal and for the cloud architects to view resource usage statistics through BI-enabled dashboards
- Intel’s Conceptual Cloud Architecture placed an emphasis on certain components within the PaaS layer (specifically the database, analytics, reporting, and the workflow) as well as VM isolation, load balancing, and High Availability at the IaaS layer.
- The physical-to-virtual sizing strategy depended very much on the type of application being run. A small workload consisting of a basic web app server could achieve a virtual machine density of 22:1 using Intel processor Xeon 5500 based servers. A medium workload consisting of SQL database servers and Java applications could achieve a ratio of 11:1, while a large app consisting of several stacked applications could yield a 5:1 ratio.
- Private clouds are recommended for those applications that require high levels of security, governance, and interoperability such as ERP and BI, while public clouds are recommended for workloads that require rapid deployment, reduced capital expenditure, and external vendor expertise such as batch jobs, and HR apps.
- Intel’s cloud vision consists of a world where the cloud is federated, automated dynamically, and client aware to secure the most optimal experience across the client continuum.
- With the increase in cloud traffic, also comes an increase in cloud storage. Data was described into four different categories, each with its own associated CAGR by 2014: Structured Data (Traditional Enterprise DB, 23.6%), Replicated Data (Backups & Data Warehouses, 24.2%), Unstructured Data (Archives, 54.8%) and Content Depots (Web, email, document sharing, and social network content, 75.6%).