Machine Learning applications on Cisco HyperFlex

Transcript

1. Chris Gascoigne March 2020 Cisco Users Group Machine Learning applications on Cisco HyperFlex
2. © 2018 Cisco and/or its affiliates. All rights reserved. Cisco Confidential 2 Presentation ID
3. © 2018 Cisco and/or its affiliates. All rights reserved. Cisco Confidential
4. © 2018 Cisco and/or its affiliates. All rights reserved. Cisco Confidential
5. © 2018 Cisco and/or its affiliates. All rights reserved. Cisco Confidential
6. © 2018 Cisco and/or its affiliates. All rights reserved. Cisco Confidential
7. © 2018 Cisco and/or its affiliates. All rights reserved. Cisco Confidential Demo
8. © 2018 Cisco and/or its affiliates. All rights reserved. Cisco Confidential “The Kubeflow project is dedicated to making deployments of machine learning (ML) workflows on Kubernetes simple, portable and scalable.” Kubeflow
9. © 2018 Cisco and/or its affiliates. All rights reserved. Cisco Confidential CCP Tenant Clusters VM VM VM Cluster 1 Kubernetes Cluster 1 Workloads Cluster 1 Ops Pod Pod Pod Kubernetes Fluentd Prometheus Kibana Hyperflex CNI Storage (HyperFlex / VMware) Networking (Nexus 9K) Compute Hardware (UCS) Hypervisor Layer (HyperFlex / VMware) Cisco Container Platform Istio M
10. © 2018 Cisco and/or its affiliates. All rights reserved. Cisco Confidential Master node VM Worker node VM Worker node VM Worker node VM Kubernetes Cluster 1 Master node VM Worker node VM Worker node VM Worker node VM Kubernetes Cluster 2 ... ... HXDP Storage Cluster VMWare ESXi VMWare ESXi VMWare ESXi HXDP Controller VM HXDP Controller VM HXDP Controller VM HyperFlex Cluster Cisco HyperFlex
11. © 2018 Cisco and/or its affiliates. All rights reserved. Cisco Confidential Demo
12. © 2018 Cisco and/or its affiliates. All rights reserved. Cisco Confidential • AI/ML is not magic • It doesn’t have to be just in the cloud • There are lots of tutorials, and content available (https://www.kaggle.com/ is a great place to start) • There are lots of tools that make your life easier • But sometimes there are too many options – I chose TensorFlow / Keras / Kubeflow because they are popular – you may make other choices. Summary