Agile Infrastructure IaaS Compute Jan van Eldik CERN IT Department Status Update 6 July 2012
Dec 13, 2015
Outline
• IaaS Compute– What are we doing?
• OpenStack Nova– What do we use?
• Deployments– What does it look like?
• Demo!
IaaS Compute overview
• Aim: Deliver a production IaaS service through a series of time-based pre-production services of increasing functionality and Quality-of-Service…
• …to manage the infrastructure more efficiently
• Different use cases:– Server consolidation– Batch computing– Disaster recovery
• This is an integration effort– As opposed to a development effort– Landb, ActiveDirectory, DataCenter operations,
…
Different Service Models• Pets are given names like
lsfmaster.cern.ch • They are unique, lovingly
hand raised and cared for• When they get ill, you nurse
them back to health
• Cattle are given numbers like vm0042.cern.ch
• They are almost identical to other cattle
• When they get ill, you get another one
• Future application architectures tend towards Cattle• .. But users now need support for both modes of working
OpenStack Compute “Nova”
• Relevant components– Nova - Compute– Horizon - Dashboard– Keystone - Identity Management– Glance - Image management
• ~6 month release cycle– Essex - April 5, 2012– Folsom - Sept 27, 2012– “Grizzly” – April 2013
OpenStack RPMs
• RPMs produced by Fedora “Cloud SIG”– Very responsive community– Standard Fedora processes– RPMs for EPEL6, Fedora 16/17
• Regular security and bug fix releases• Minimal delay between OpenStack releases and RPM
availability
• Note: RedHat joined the OpenStack Foundation in April, but they have not yet revealed any strategy
Configuration, Images, …
• Configure using Puppetlabs modules– Recently ported to Fedora and RHEL6
• With CERN help
– Released last week• Images created with Oz– Part of (RedHat sponsored) project Aeolus– SLC6, SLC5 “server” images– Contextualization w/ cloud-init
• Interpreting user-specified data
– WIP: Windows images• Use GlusterFS– Provide block storage, VM live-migration
CERN deployments
Guppy• First pre-production, released June 1st
• Target audience: AI project members• OpenStack Essex on Fedora 16, KVM only • Used for functionality tests– Live-migration, resize, …
• Limited integration in Cern infrastructure– single IP service, no Cern accounts
• School of guppies to be deployed as an AI testbed
CERN deployments
Hamster• Scheduled for Sept 1st
• AI project members, early adopters• OpenStack Essex on SLC6 • Proper integration in Cern infrastructure– multiple IP services, user defined hostnames– Accounts and roles from Active Directory– Window images
• Demonstrate VM backup• Deploy Hyper-V hypervisors
CERN deployments
IaaS Compute v 1.0• Production release, scheduled Q1 2013• OpenStack Folsom on SLC6 • To be deployed in Meyrin and Budapest DCs– Batch worker nodes– Service nodes, selected pets (Meyrin only)
' 12 Apr2012 Jun Aug Oct Dec Feb
2013 Apr ' 13
Today "Grizzly"April 2013
Iaas Compute 1.0Q1 2013
FolsomSept 27, 2012
HamsterSept 1, 2012
GuppyJune 1, 2012
EssexApr 5, 2012
• OpenStack is rapidly maturing• With a vibrant eco-system of developers, distributions,
deployers• We are delivering a series of pre-production services…
• Adding functionality…• Scaling out…• Integrating into IT infrastructure…
• For cattle and pets…• Aiming for a production service in Q1 2013