70 likes | 80 Views
Embrace marginalization and evolve building blocks for innovation in response to user needs and regulations. Explore the importance of resiliency and telemetry in detecting and preventing abuses. Learn how to navigate dependencies for successful innovation.
E N D
Campus Drivers Jon Moore Ron Hutchins Kevin Miller Mark Poepping
Drivers? Not us • We’re Responders – sometimes “first” • Responding to: • User ‘needs’ – each application interaction is an overlay network from the app/user PoV • Vendors – metamorphosing building blocks • Regulations – left field and buzzword • By expediency or soapbox? • Operationally: answering the phone
Lesson1: Marginalization is inevitable… • Good general building-block models enable innovation in areas unexpected. Often leads to a few… • high-value ideas that people *really* start to like (willing to pay for), leading to… • specialization that makes that cool thing really efficient, leading to… • marginalization where the general model is no longer a value for a majority, meaning.. • violation of generality is the expedient thing, making innovation increasingly problematic
Lesson2: Need to appreciate and keep working both ends • If we don’t embrace marginalization, we become irrelevant • If we don’t rework our building blocks, we lose ability to change • Oh by the way, we don’t own all the building blocks (personal lambdas, cellular, overlays - see above)
More Lessons: • Resiliency masks fixable problems • Allows us to be lazy • If it can be abused, someone will • If you can’t prevent all abuse you need to be able to detect it. • If you have to detect anyway, then you can optimize prevention and detection, maybe you don’t need so much prevention • We have very little by way of telemetry to support detection in today’s ‘interactions’
Telemetry? Data to help with… • “Please just tell me” • What’s my machine doing? • Weather Report or “Eyewitness Account” • Performance of the next packet, or the last packet? • So what the heck *really* happened • Who says what happened • Digital forensics
Other Telemetry • Dearth of useful diagnostic (mgmt) data • Data useful in application stovepipes, but.. • Interdependent infrastructures need to share too • Among layers/peers in technology • Among layers/peers of organizations • “Dependency in seemingly unrelated events” • Sharing is hard, unpredictable, changing, and ultimately Essential