Reflections on the Cloud Expo in Silicon Valley and How Do I Know My Apps are Working in the Cloud?
Cloud Expo was indeed a very interesting juxtaposition of people espousing the value of cloud and how their stuff is really cloudy. You have a group of presenters and expo floor booths talking about their open API and how that is the future of cloud. Then you have the other camp that tells us how their special mix of functions is so much better than that. All of this is a very interesting dialog. APIs are indeed very important. If your technology is indeed a cloud operating model then you must have an API. Solutions like Cisco’s Intelligent Automation for Cloud rely on those APIs to orchestrate cloud services. But APIs are not the end all. The reality is that while the cloud discussions tend to center on the API and the model behind that API, the real change enabling the move towards cloud is the operating model of the users who are leveraging the cloud for a completely fresh game plan for their businesses.
James Urquhart’s recent blog: http://gigaom.com/cloud/what-cloud-boils-down-to-for-the-enterprise-2/ highlights that the real change for users of the cloud is modifying how they do development, test, capacity management, production operations and disaster recovery. My last blog talked about the world before cloud management and automation and the move from the old world model to the new models of dev/test or dev/ops that force the application architects, developers, and QA folks to radically alter their model. Those that adopt the cloud without changing their “software factory” model from one that Henry Ford would recognize to the new models may not get the value they are looking for out of the cloud.
At Cloud Expo I saw a lot of very interesting software packages. Some of them went really deep into a specific use case area, while others accomplished a lot of functional use cases that were only about a inch deep. As product teams build out software packages for commercial use, they have a very interesting and critical decision point that will drive the value proposition of the software product. It seems to me that within 2 years, just about all entrants in the cloud management and automation marathon will begin to converge on a simple focused yet broad set of use cases. Each competitor will be either directly driving their product to that point, or they will be forced to that spot by the practical aspects of customers voting with the wallets. Interestingly enough, this whole process it drives competition and will yield great value for the VP of Operations and VP of Applications of companies moving their applications to the cloud.
Operation and Application teams intersect in very clear objective, deploying cloud applications and keeping them available and high performing. Zenoss’s Kent Erickson http://blog.zenoss.com/2011/11/cisco-zenoss-clear-it-up-at-cloud-expo-2011/ and I posed an interesting question to the crowd at the Cloud Expo: How do we know a Cloud Application and the service it provides to end users is meeting the requirements of acceptable performance and availability? Let’s assume that your favorite Cloud Automation software vendor provisioned a set of physical and virtual servers running a composite application with a set of network services all supported by your favorite storage vendor(s). After all the provisioning and orchestrated API calls, how do we assure that the service is running well? When there is a problem, how do I separate out all the alarms going off? Zenoss (and other vendors) have a pretty interesting take on monitoring the performance and viewing that service topology, as well as reducing those 100’s of alerts operations tends to get when there are one or two key issues in the infrastructure or application. The follow image shows a service topology tree of a fairly complex composition application running in the cloud and how the key red zones can yield a warning for the application stakeholders and drive to efficient root cause….
Provisioning cloud infrastructure for applications will require this class of software to drive service assurance for Cloud services. Cisco’s Intelligent Automation for Cloud has integrated with many monitoring and service assurance packages such as from Zenoss, Watch4net, and CA’s Nimsoft to name a few. I have observed that the customers I talk to every day have their favorite products and it is even clearer to me that customer choice and integration is key to a successful deployment in the “real world”.