Overview
Explore deterministic storage performance in OpenStack and Ceph systems, emulating AWS's capacity-based Quality of Service (QoS) approach. Dive into how Ceph storage pools and Cinder Volume Types can replicate Amazon's Magnetic and Provisioned IOPS Elastic Block Store (EBS) functionality. Analyze performance graphs to determine when size-based or fixed limits are more suitable, depending on the underlying storage media. Examine block storage trends in public clouds, OpenStack frameworks, and Cinder drivers. Investigate universal scalability models, SSD-based clusters, and scaling dimensions. Learn about capacity-based limits and their impact on performance. Discover the tools and gaps in OpenStack for implementing QoS with Ceph storage, enabling you to provide familiar public cloud modalities in private cloud environments.
Syllabus
Intro
WHY DOES IT MATTER?
PROBLEM STATEMENT
BLOCK STORAGE IN A PUBLIC CLOUD
THE AWS WAY
EBS Volumes: an example
THE GOOGLE WAY
OPENSTACK FRAMEWORK TRENDS
OPENSTACK CINDER DRIVER TRENDS
BLOCK STORAGE WITH OPENSTACK
UNIVERSAL SCALABILITY MODEL
SSD BASED CLUSTERS
SCALING DIMENSIONS
CAPACITY BASED LIMITS
SUMMARY OF RESULTS
CEPH + OPENSTACK QOS
OPENSTACK TOOLS AND GAPS
Taught by
Red Hat