DataArt Innovation Showcase Immutable Diskless Pool Demo
Download as PPTX, PDF1 like398 views
The document discusses DataArt's offerings and the implementation of an immutable diskless compute pool within OpenStack. It highlights the advantages of this approach, such as reduced points of failure, faster deployment, and seamless live migration capabilities. The document outlines the necessary setup steps and the architecture that supports this technology.
DataArt Innovation Showcase Immutable Diskless Pool Demo
- 1. New York USA London UK Munich Germany Zug Switzerland New York USA London UK Munich Germany Zug Switzerland TAD Summit 2017 Immutable Diskless Pool Demo
- 2. Introducing DataArt: Global Technology Consultancy 22 Global Locations: • NYC • London • Switzerland DataArt’s Core Offers • Controlling Costs via Managed Services • Building New Products and Services • Modernizing and Re-engineering Legacy systems • Consulting on New Technology Approaches Providing On-Demand IT In operation 20 years Consultants & engineers 2500+ Staff Turnover <7% Return clients 95% Development Hours 15+ million Successfully completed projects 1700+ Finance Betting and Gaming TelecomMedia & Entertainment IoTHealthcare & Life Sciences Travel & Hospitality
- 3. Immutable Diskless Compute Pool Technologies
- 4. The default approach to instantiate boot volumes in OpenStack Ocata (and newer) is using Cinder backed block storage. • When compute pool is growing, the risk of failure of particular node is increasing along with effort required to maintain the compute pool and keep it up to date • In case if Cinder is deployed with CEPH or with any supported clustered SAN, it is possible to implement live migrations between different compute nodes in the pool Immutable Diskless Compute Pool Background
- 5. Since all VM volumes are stored remotely, only a small number of files is written to the operating system root volume (usually configuration files, DHCP leases and libvirt XMLs). • It’s possible to create an image with all required software for compute node (nova-compute, neutron, openvswitch), then boot it via PXE to RAM and synchronize configurations files • Use BTRFS in-RAM along with provisioning scripts • This task is an excellent candidate for automation using CI and Containers Immutable Diskless Compute Pool Solution
- 6. •Immutable Diskless Pool Demo
- 7. 1. Setup PXE Server, Key Broker, Jenkins and Dashboard on the master node 2. Add diskless compute nodes to the pool 3. Deploy OpenStack Pike via Ansible and dynamic data source to Docker Containers across all the nodes from the Docker private registry, built by CI using Kolla project 4. Create VM on the diskless compute node 5. Emulate the compute node’s failure – switch off the compute node in the VirtualBox 6. Migrate VMs to another compute node using CLI and make sure that it works Immutable Diskless Compute Pool Showcase
- 8. • OpenStack controllers in HA • Neutron gateway • Clustered Cinder on Ceph • Jenkins that builds OpenStack Docker images with the latest security updates • TFTP and NFS node for serving PXE images • Key service with Ansible API for RSA key distribution and further configuration • Diskless pool for nova compute Immutable Diskless Compute Pool Architecture
- 9. • Vendor-neutral IaaS, can use existing open-source or commercial SAN • Can be built from sources or binary packages • Better TCO due to better drives utilization • Faster deployment/upgrade using private image registry and In-RAM provisioning • Less points of failure because compute pool doesn’t store any significant data • Instant migrations in case of any failure (nova host-evacuate) • Secure deployments without internet access, unique RSA fingerprints for each start of node Immutable Diskless Compute Pool Advantages Over Disk-Based Deployment
- 10. •Thank You!
Editor's Notes
- #3: I include DataArt’s corporate offers as you could win business with a telco as an enterprise – not as a telco, so I think its important to state the corporate positioning – also provides consistency to your website.