CloudCamp @ CodeCamp, Silicon Valley 2009-10-04

From the one-minute summaries at the wrap-up.

Maintaining a Cloud

 * capacity-planning
 * solved problem in industrial settings
 * virtual capacity planning

SQL Azure

 * what's in & what's not
 * 1G & 10G

Intro to Cloud Computing

 * evolving to solve business problems
 * operational efficiency
 * platform as a service (from developer perspective)
 * infrastructure as a service (from IT perspective)
 * software as a service (from business perspective)

Security and Privacy

 * policy management between enterprise and cloud
 * identity
 * authentication
 * provisioning
 * cloud exposes fundamentals of security policy - if you were broken before, it'll be even more obvious on the cloud

Writing Cloud-native Applications

 * learn what services are already on the cloud
 * write to those
 * simplecloud.org
 * memcached is a super temp fix for database overload problems

Benchmarking in the Cloud

 * interest in build an open-source, community-based yardstick for cloud computing resources
 * need benchmarking in the cloud
 * need a lot of sampling to accomodate time and service based variability
 * needs to be community-based so there isn't vendor bias

Business Case of Cloud Computing

 * if your apps are quick, easy and cheap, good for cloud
 * apps that need to be elastics are good for the cloud
 * SLAs are still not well-developed, but should develop as market develops

CloudBursts

 * needs to become standard operating procedure
 * be balanced all the time, part in, part out
 * do a little now to be ready for scaling or disaster recovery when you need it
 * multiple internal clouds, share resources between
 * best practices become more imporant - bad practices are worse in the cloud

Adapt Enterprise Services to Cloud

 * offer as a service, but a different service so as not to disrupt existing users

Standards
(missed this summary)

Analytics and Large Data

 * made distinction between HPC (high-performance computing) and cloud computing
 * cloud computing needs to work for regular programmers
 * need to make parallel computing look like regular computing (many CPUs like one)
 * need to

Startups and Cloud Computing

 * have appropriate product or service to run on cloud
 * failing fast good for a startup
 * if app sucks on-premise, it will suck more on the cloud
 * choice of cloud vendor very important, influences your design
 * infrastructure services less risky than platform services, because you can move more easily