We often talk about organizations running hybrid workloads on hybrid platforms using hybrid architectures. And we preach about hybridization as though it’s a voluntary evolutionary step that organizations should take. But we rarely consider the hybrid organization — the user of a platform that is, in and of itself, a conglomerate or an amalgam of various classes of users in differing capacities scattered throughout the planet.
The Institute for Health Metrics and Evaluation at the University of Washington is, on its face, a single research facility that just happens to claim the largest repository of human health data anywhere in the world. But that repository is by virtue of a unique organizational structure that flips the whole question of hybridization on its ear (assuming that such an organization keeps its ears in the same place). From the perspective of the applications it uses, IHME is a hybrid organization. A few hundred research facilities worldwide, all of whom have some responsibility for gathering health care data, also share it.
Yet the platform itself is actually not the cloud. In fact, it’s a centralized data center, on the grounds of the University. Its use of Univa’s Navops, a multitenant adaptation of the open source Kubernetes container orchestration engine, enables all these facilities to use the centralized platform as though it were the cloud. It ensures that UW’s IT team manages governance and security, which many countries in which its users reside require.
But this special edition of The New Stack Makers podcast is not just about abstract infrastructure and multitenancy. This is a use case whose results lead to insights about the way countries and cultures can better treat their own people — helping them avoid death and improve their quality of life. In a world where good health care often stands precariously perched on the precipice, the sharing of information could become a lifeline for tens of thousands.
Listen now to this story of Navops, the IHME, and the real reason we work with infrastructure and automation every day.
3:57: How IHME improves citizen health via creating a health database.
7:22: Exploring what types of applications are necessary for IHME and its partners to make sense of that data.
11:00: How Univa’s NavOps and Kubernetes helped IHME’s partners manage their infrastructure.
23:49: Academic workloads and policy as it applies to academic and research workload versus a commercial workload.
29:45: How IHME came to embrace NavOps specifically and apply its scheduling techniques to those purposes.
36:57: Projecting IHME’s resource consumption needs, scheduling, and infrastructure management needs three years from now with NavOps.