2. Agenda
1.Before Openstack
2.Profile of Openstack
3.OpenStack Architecture and Components
4.OpenStack, SDN & NFV in Telco Environments
5.CERN Cloud Architecture
6.China Mobile
7.AT&T
2
3. CONVENTIONAL DATA CENTRE
❖ Known for having a lot of hardware that is, by current standards at least,
grossly underutilized
❖ All the hardware and their software are usually managed with relatively
little automation.
❖ Very hard to find the right balance between capacity and utilization
❖ Variety of Applications
3
4. Manual Intervention
❖Problem: Network Integration, Monitoring, Setting up high availability and
Billing
❖Not hard to automate
❖Existing automation frameworks like Puppet, Chef, JuJu, Crowbar or
Ansible are sufficient to automate the whole process
❖Virtualization:
• Deploying a new system is fairly easy via provisioning a new VM
• Yet, many things need to be done manually
4
5. Advantages of Automation
❖Cloud provider’s task: provide customers with resources and ensure it is
enough any time
❖Cloud provider adds more resources when needed
❖Automation can facilitate flexibility of the new resources in terms of
network integration, monitoring, etc…
❖Users can start and stop VM in clicks
5
6. Automation
❖Authorization Scheme: that matches clients’ requirements e.g. managers
stop/start VM while Administrators can add/remove VMs
❖Image Management: upon creating new VMs, clouds need pre-made
images so that users do not have to install OSs by themselves
❖Resources Management e.g. processing power, storage, and network
❖Existing cloud solutions: OpenNebula by NASA, OpenQRM, Eucalyptus
and OpenStack
6
8. Introduction
❖An open source cloud platform.
❖Controls large pools of compute, storage, and networking
resources throughout a datacenter.
❖All managed by a dashboard that gives administrators control
while empowering their users to provision resources through a web
interface.
8
19. Nova
❖Provides compute as a service
❖The main part of an IaaS system
❖It is designed to manage and automate pools of computer resources
❖Compute's architecture is designed to scale horizontally
19
21. Nova - Components
❖nova-conductor: Provides database-access support for Compute nodes
❖nova-consoleauth: Handles console authentication
❖nova-novncproxy: Provides a VNC proxy for browsers
21
22. Nova API
❖nova-api is responsible to provide an API for users and services to
interact with NOVA
22
25. Keystone
❖Keystone is the identity service used for Authentication
❖Set of assigned user rights and privileges for performing a specific set of
operations
❖A user token issued by Keystone includes a list of that user’s roles.
Services then determine how to interpret those roles
25
28. Glance
❖The Glance project provides services for discovering, registering, and
retrieving virtual machine images.
❖Glance has a RESTful API that allows querying of VM image metadata as
well as retrieval of the actual image.
28
30. Cinder
❖Architected to provide traditional block-level storage resources to
other OpenStack services
❖Presents persistent block-level storage volumes for use with
OpenStack Nova compute instances
❖Manages the creation, attaching and detaching of these volumes
between a storage system and different host servers
30
33. Swift
❖ A distributed object storage system designed to scale from a single
machine to thousands of servers
❖ optimized for multi-tenancy and high concurrency
❖ •ideal for backups, web and mobile content, and any other
unstructured data that can grow without bound.
❖ Swift provides a simple, REST-based API
33
36. Ceilometer
❖OpenStack Telemetry provides common infrastructure to collect usage
and performance measurements within an OpenStack cloud.
❖ Its primary initial targets are monitoring and metering
❖collect data for other needs.
❖Ceilometer was promoted from incubation status to an integrated
component of OpenStack.
36
37. Ceilometer Workflow
37
❖Collect from OpenStack components
❖Transform meters into other meters if necessary
❖Publish meters to any destination (including Ceilometer itself)
❖Store received meters and read them via the Ceilometer REST
API
39. Trove
❖OpenStack Database as a Service
❖high performance ,scalable and reliable
❖relational and non-relational database engines
❖Trove was promoted from incubation status to an
integrated component of OpenStack.
39
41. Sahara
❖OpenStack Hadoop as a Service
❖Aims to provide users with simple means to provision a Hadoop cluster
by specifying several parameters
❖ Sahara was promoted from incubation status to an integrated
component of OpenStack.
41
43. Manila
❖OpenStack File Share Service
❖Provides coordinated access to shared or distributed file systems.
❖Manila was officially denoted as an incubated OpenStack program
during the Juno release cycle.
43
46. Neutron
❖Network as a Service (NaaS)
❖Provides REST APIs to manage network connections for the
resources managed by other OpenStack Services
❖Complete control over the following network resources in
OpenStack(Networks, Ports and Subnets)
❖Build complex network topologies
❖Limited L3 functionality (IP tables rules at host level)
46
51. Neutron Components
❖Neutron Server
• Implement REST APIs
• Enforce network model
• Network, subnet, and port
• IP addressing to each port (IPAM)
❖Plugin agent
• Run on each compute node
• Connect instances to network port
❖Queue
• Enhance communication between each
• components of neutron
❖Database
• Persistent network model
51
52. Neutron Components
❖DHCP Agent (*)
• In multi-host mode, run on each compute node
• Start/stop dhcp server
• Maintain dhcp configuration
❖L3 Agent (*)
• To implement floating Ips and other L3 features,such as NAT
• One per network
52
57. Transformation of Carriers
Business Model
❖Complex and expensive infrastructure
• Challenging to operate and maintain
• slow rolling out of new services
❖Cloud-based Model
• Always-on services
• Affordable
• Reliable
• First attempt: Cloud RAN
57
58. Production Ready: NFV with
OpenStack
❖Deployed on cost effective Commercial Off-The-Shelf (COTS) hardware
❖Based on Open Source Software
• Can be easily adapted to any customization
• Community Driven
❖Standard APIs
❖Software-managed High Availability (HA)
❖AUtomated Deployment
❖Virtualized Infrastructure
• Scalable
• Upgradable
• Optimizable
• Modular
• Customizable
58
59. Production Ready: NFV + SDN +
OpenStack
Software Defined Components
Resilient and Reliable
Flexible and Extensionable
Optimized for Performance
Secure
59
60. Carrier Benefits
❖Network Operations Benefits
• Ease of automation
• Increased Deployment Agility
• Visibility with monitoring and alerting
• Reliable
• Self Healing
• Highly Available
❖Cost Benefits
• Multi-tenant
• Flexible
❖Secure at Each Layer of the stack
60
61. OPNFV + OpenStack
❖OPNFV is a carrier-grade, integrated, open source platform for NFV
products and services
• widespread collaboration across many telco
• uses OpenStack as Virtualized Infrastructure Manager
❖Telco running NFV implementations includes AT&T, China Mobile,
Orange, SK Telecom and Telecom Italia
61
63. OPNFV + OpenStack
“We are fully committed to open networking and open source including our
work with OPNFV and OpenStack” - Alex Zhang, Principal Architect, China
Mobile
“To keep up with the exponential growth of its network, AT&T is deeply
committed to using open source networking technologies in our software-
centric network. As we work to virtualize more of our network and implement
a common infrastructure for VNFs, OpenStack and OPNFV will become
important parts of our technology stack” - Margaret Chiosi, Distinguished
Technical Architect, AT&T
63
64. Case Study: vCPE
❖vCPE: Virtual Customer-Premises
Equipment
❖Existing Solution:
• Edge networking devices are
standalone nodes
• Provide advanced services (QoS,
Dynamic Routing, NAT…)
• Complex software, prone to failure
• Cheap Hardware, prone to failure
• Cannot be easily Upgraded or serviced
64
65. Case Study: vCPE
❖ Apply SDN, NFV and OpenStack to
the network
❖Move Control Plane to core network
❖Keep Data Plane at customer
premises with additional
microservices
❖Benefits
• Reduce CAPEX and OPEX
• Improve service agility
• Deliver personalized services
• Transition to SaaS-based business 65
67. What is CERN?
❖European Organization for Nuclear Research
❖Founded in 1954
• 21 state member
• other countries contribute to experiments
❖Situated in the Swiss-French border
❖Do fundamental research
67
68. CERN’s Large Hadron Collider
❖Biggest machine in the world
❖27km Tall - 175m underground
❖Accelerate 2 particle beams
traveling near speed of light
❖Beams collide in 4 different points
of detectors
❖Detectors are 100 MP digital
cameras 14000000 times in a
second
❖Generates 25 PetaBytes per year
• Estimated 400 PB / year by 2023
68
69. CERN Data Centers
❖2 Data Centers; one in Geneva and another in Budapest
❖Data Centers are managed by OpenStack
❖190k+ cores on 5000+ compute nodes running KVM and Hyper-V
❖16000+ VMs
❖~160 PetaBytes stored at CERN
❖June-August 2016: recorded > 0.5 PB
❖2400+ Images, 2000+ Users, 2500+ Projects
69
70. ❖CERN deployed OpenStack in 2013
❖Nova, Keystone, Glance, Heat, Horizon, Ceilometer, Rally
❖26 Nova cells
• Single endpoint to users
• Scale transparently between Data Centres
• Availability and Resilience
• Isolate different use-cases
❖HA only in the top cell
❖2 Ceph instances
• A free-software storage platform, implements object storage on a single distributed
computer cluster
OpenStack at CERN
70
73. Nova - Cell Scheduling
❖ Different Cells has different hardware, configuration, hardware,
Hypervisor type
❖Cell Scheduling is the process to schedule operations according to cell
capabilities e.g. hardware, availability
❖Schedulers filters to use these capabilities
❖It enables mapping projects to cells and restrict cell usage according to
project type
73
74. Nova-Network? in CERN
❖CERN uses Nova-Network instead of Neutron
• An OpenStack networking module before Neutron
• Deprecated
• Better than Neutron in some use cases
• Planned to migrate to Neutron
❖ Migration to Neutron, No Use of:
• SDN or tunneling
• Only provider networks
• Flat networking. VMs directly connected to the real network
• Floating IPs
• DHCP or DNS Neutron services. Already have infrastructure
74
75. Keystone in CERN
❖Two different keystone
infrastructure
• Exposed to users
• Dedicated to Ceilometer
❖Keystone nodes are VMs
❖Integrated with Active Directory
❖Project lifecycle
• ~200 arrivals/departures / month
• Users subscribe to the cloud
service
• Limited Quota of personal projects
• Shared projects created by
request
75
76. Glance in CERN
❖Uses Ceph backend in Geneva
❖Glance Nodes are VMs
❖Two sets of nodes: Exposed to user and Ceilometer
❖No Glance image cache
76
77. Cinder in CERN
❖Ceph and NetApp backends
❖Extended list of available volume types (QoS, Backend, Location)
❖Cinder nodes are VMs
77
80. Who is China Mobile?
❖One of the world’s largest telecommunication service providers
• Huge network scale
• Huger customer base
• Large market value
❖At end of 2014
• 800M+ subscribers
• 2.2M+ base stations
• Covered more than 99% of the population of PRC
80
81. NovoNet
❖Vision for the next-generation
network by 2020
❖High-quality intelligent network
❖Based SDN and NFV
81
82. NovoNet
❖Firstly deploy in Cloud Data Centers
and Packet Transport Networks
(PTN)
❖Using OpenStack and
OpenDayLight
❖Goal: Build out several enterprise
service offerings under NovoDC
including a virtual private cloud
82
85. Who is AT&T?
❖American multinational telecommunications
❖Already handling 114 PB a day of data
❖By 2020, At&T network is expected to jump 10 folds
❖Global Customers
85
86. AT&T Future Network
❖Move 75 percent of its network infrastructure to the cloud
❖Make greater use of software-defined networking (SDN) with
OpenDaylight and Open vSwitch
❖Goal: Reduce deployment times for cloud "zones" from months to days
❖Use OpenStack tools to develop an end-user "resource manager"
❖Working on AT&T Integrated Cloud (AIC)
• 74 AIC zones in 2015
• 105 AIC zones in 2016
• 1000+in 2020
• All running OpenStack
86
87. References
❖http://openstack.org
• Tokyo Summit 2015 https://www.openstack.org/summit/tokyo-2015/
• Austin Summit 2016 https://www.openstack.org/summit/austin-2016/
• Barcelona Summit 2016
• https://wiki.openstack.org/
❖http://linux.com
• Linux Foundation Blog
https://www.linux.com/blog/learn/chapter/openstack/essentials-openstack-
administration-part-1-cloud-fundamentals
❖https://www.opnfv.org/
❖https://www.sdxcentral.com/cloud/open-source/definitions/
❖https://www.opendaylight.org/news/user-story/2015/11/china-mobile-
builds-next-generation-network-opendaylight
❖http://about.att.com/innovationblog/openstack_superuser
87