As enterprise cloud usage grows, operators struggle to maintain a low-cost, monolithic infrastructure while satisfying users’ increasingly-specialized requirements and usage demands.
Most cloud operators understand the value of OpenStack (Nova) flavors in mitigating this issue, but some struggle to effectively utilize this feature because:
- They lack an adequate process and automation for flavor lifecycle management
- They look at tenant requirements in a siloed fashion and lack a big picture focus on the larger need to offer specialized, but standardized configurations
- On the other extreme, they make the mistake of trying to standardize too much without permitting the degree of specialization needed for different use-cases (the 'everything must be the same!' mantra)
- They have sub-optimal flavor geometries that do not maximize use of available resources
In this session we'll look at how to design flavors to prevent these problems.
Attendees will learn about:
- Flavor terminology, semantics, and conceptual breakdown
- What can go wrong with flavors (things to avoid, lessons learned)
- Best practices and industry standards around planning and managing flavors
- How to manage/mitigate cost of introducing additional flavor offerings
- Gain an appreciation for some of the design tradeoffs for different optimization decisions
- How to plan for and account for legacy workloads alongside cloud native workloads with respect to flavors
- How flavors impacts and links together many disciplines and stakeholders (architecture/planning, operations, end-users, automation, etc)