Sonian has been innovating in cloud compute environments for over 3 solid years. By Internet-speed time standards we’re considered “seasoned” veterans in bending the cloud to work our magic. We understood from the early days that Software as a Service powered by cloud computing is very different than SaaS powered by legacy data center designs.
It’s mentally and physically “freeing” not to have to worry about the underlying physical infrastructure that supports our software stack, yet there are other challenges in the cloud that need rapid innovation to ensure we are always making the most out of what the cloud is really good at: on-demand CPU. Our whole software stack design is built on the premise of elastic CPU: for scale, redundancy, and cost effectiveness.
In the cloud, where in theory all one needs to think about is software, system administration issues are much less about hardware, storage and networking and more about OS and automation. Which means the skill set required to be a great “cloud-enabled” sys admin is a blend of programming and OS skills, much different from managing hardware, storage systems and networking gear.
To be truly successful in the cloud requires a commitment to 100% automation. All aspects of system administration need bullet-proof automation. At Sonian we are investing heavily in automation at all levels of the software stack, which means no matter which function (web server, database server, index server, etc.) is able to launch from “bare metal cloud” and configure to the role automatically. All this is possible with the new skill set of “developer sys admin” (what we’re calling DevOps), and is emerging as a key requirement to successful cloud deployments.
A lesson we have learned is to pair-up a “DevOps” person with the traditional developer so that automation is happening all throughout the coding process, and not at the end of the development cycle. Don’t short-cut your automation work and incur technical debt that will need to be repaid.