Transcript NFV_SDNx - Computer Networks

Agenda





2
Who is Ixia
Overview of SDN and NFV technology
SDN
NFV
NFV validation issue
 Publicly traded: XXIA
(NASDAQ)
Solutions and services
 End-to-end visibility into the performance
 Security of pre-deployment and production networks
Sales & Support
R&D
MFG / Operations
Corporate
R&D, Operations, Sales
Partner Sales & Support
iSimCity Network Simulation Facility
 Global company
 1,800+ employees
 2,500+ loyal customers
The Ixia Solution
Application Performance
• Validate network functions in pre and
post deployment environments
• Optimize traffic across multiple
infrastructures
• Ensure service level agreement (SLA)
compliance
• Load balance between functions for
optimal application performance
Security Resilience
• Simulate real world attacks and
malware threats
• Deliver real time threat intelligence for
proactive protection
• Harden the resiliency of critical
security devices
• Train cyber warriors for attack
readiness
GLOBAL ADOPTION
NEMs
Service Providers
Enterprise
Government
Overview of SDN and NFV technology
What is virtualization ?
Virtualization, in computing, refers to the act of creating a virtual (rather than
actual) version of something, including but not limited to a virtual computer
hardware platform, operating system (OS), storage device, or computer
network resources.
COMPUTE, STORAGE and NETWORK
How has it transformed the way we do business?
Server Virtualization (Compute
and Storage)
• Pioneered by VMWare
•
Saved organizations Billions in capital
expenses through Server
Consolidation
• Adopted by Amazon AWS
•
•
•
•
•
Hosts its own retail website
(www.amazon.com) on AWS
Netflix uses AWS for streaming and
delivery of content
Has revolutionized how we access our
data
Enables virtualization across different
manufacturers products
Emergence of Cloud Storage services like
Box, Dropbox
Virtualization of the Network had been left behind
Why problems exist with current networks and systems ?
Current model:
• Service providers want
Applications
Slow to deploy
Management
Non-standard
interfaces, high OPEX
cost, difficult
provisioning
Network Infrastructure
high CAPEX cost, high complexity,
proprietary, protocol soup
• To deliver & monetize on new
services
• To manage and quickly tune the
network to new apps/services
• All without the overhead of new
CAPEX or network overhauls
What do they need ?
The industry needed a new networking
paradigm that lends itself to creating
higher level application-aware
abstractions on top of a physical
network
High value, centralized
applications, easy to deploy,
ability to monetize
They need
• The ability to build an architecture which
makes the innovation and feature
adoption velocity in the networking
industry comparable with the software
industry.
• Reduce the total cost of ownership by
eliminating vendor lock-ins
• Reduce system complexity and brittleness
by eliminating proprietary software
Common/Open API to
manage and provision
multi-vendor
equipment
Lower Cost
Current Networks - Challenges
• Difficult to optimize
•
FEATURE
FEATURE
• Known issues
•
OPERATING SYSTEM
SPECIALIZED PACKET
FORWARDING HARDWARE
FEATURE
FEATURE
FEATURE
Difficult to introduce new revenue
generating services and optimization and
adding new services is difficult
FEATURE
OPERATING SYSTEM
OPERATING SYSTEM
SPECIALIZED PACKET
FORWARDING HARDWARE
SPECIALIZED PACKET
FORWARDING HARDWARE
Issues like security, robustness,
manageability and mobility have not
been addressed so far
• Capital costs
•
•
Network capital costs have not been
reducing fast enough
Operating costs have been growing
• Difficult to customize
FEATURE
FEATURE
OPERATING SYSTEM
SPECIALIZED PACKET
FORWARDING HARDWARE
Network of vertically integrated, closed proprietary switches
•
Difficult to create customized cost
effective solutions on closed, proprietary
solutions
Software-Centric Network
FEATURE
FEATURE
OPERATING SYSTEM
SPECIALIZED PACKET
FORWARDING HARDWARE
FEATURE
FEATURE
FEATURE
FEATURE
OPERATING SYSTEM
OPERATING SYSTEM
SPECIALIZED PACKET
FORWARDING HARDWARE
SPECIALIZED PACKET
FORWARDING HARDWARE
FEATURE
FEATURE
OPERATING SYSTEM
SPECIALIZED PACKET
FORWARDING HARDWARE
Network of vertically integrated, closed
proprietary switches
Separation of control and data plane
Open Interface between control and data
plane
Open Interface to the control plane
Network control and management features in
software
Software Network – The Benefits
• Infrastructure Savings
•
•
No need for specialized hardware
Ordinary software running on ordinary
operating systems on ordinary servers
• Flexible and customizable
•
New services or customized network
behavior can be enabled quickly
through software
• Slice the network
•
Network slicing and virtualization make
it easier to experiment with new
capabilities in isolated slices of the
network without impacting other parts
of the network
• Enables Network Virtualization
Separation of control and data plane
Open Interface between control and data
plane
Open Interface to the control plane
Network control and management features in
software
•
This significantly improves efficient use
of network resources with multiple
customers and services.
SDN and NFV relationship
• SDN and NFV do not need to
be deployed together
• Mutually exclusive, but
complimentary
• Most commercial
deployments will use a
combination of SDN and NFV
• NFV service chaining can
be accomplished using
SDN (SFC)
• SDN controllers can be
virtualized in a NFV
environment
Is it real ?
Infonetics 2014 Survey : SDN and NFV Strategies: Global Service Provider Survey
•
•
•
•
•
29% of participants are currently implementing SDNs
52% plan to evaluate SDNs by the end of 2015
97% of participants plan to deploy SDN
93% of participants plan to deploy NFV
Business vE-CPE VNF rated as the #1 use case overall for NFV in 2014–2015
What does this mean for NEMs ?
•
•
•
•
•
Need to find a way to differentiate their hardware
Develop new protocols/standards to differentiate
Ensure comparable performance with legacy protocols
Ensure comparable performance with virtual variants
Makes marketplace more competitive with COTS hardware
What does this mean for everyone?
• Everything old is new again
•
•
Need to move from well known, tried and tested running on custom HW to new untested,
SW based solutions running on off-the-shelf hardware
Need to ensure predictable and comparable levels of performance, scale ,availability and
security
• Implementation of a new level of management
•
Need to way to manage all the virtual functions and networks (Orchestration)
• Need support for Virtual Network Functions (VNFs)
•
•
Custom developed VNFs could delay function availability
Off-the-shelf cloud apps/functions are largely untested
• Hold everyone to the same standard
•
Need more standardization of NFV standards
SDN
Start of SDN idea
• Martin Casado
• OpenFlow begun at Stanford University in 2008
• By December 2009, Version 1.0 of the OpenFlow
switch specification was released
• OpenFlow has been managed by the Open
Networking Foundation (ONF)
SDN (Software Defined Networking)
Software Programmability of
the Network
• Designed to simplify traffic
management
• SDN allows a controller to
define how packets are
forwarded by networking
elements
• Its about the separation of
control and data planes
within switches/routers
• 3 distinct SDN features
• Separation of the control
plane from the data plane
• A centralized controller and
view of the network
• Programmability of the
network by external
applications
Business
Applications
Application
Layer
App
App
App
App
SDN
Controller
API
Control Plane
Control
Layer
Control and Data
Plane Interface
(E.g. OpenFlow)
Forwarding
Layer
Data
Plane
Data
Plane
Data
Plane
Network
Device
Network
Device
Network
Device
Enables Network Virtualization
HP SDN App Store
SDN Building Blocks – How It Happens
Separate control and data plane
SDN offers
• Elasticity
• Simplicity
• Agility
• Value added services
Openflow
What is OpenFlow?
OpenFlow is a communication protocol between a Controller and
an OpenFlow enabled Switch providing a standard API to control
packet forwarding
Background
• started at Stanford, now IPR controlled by the Open Networking
Foundation (ONF)
• 90+ member companies
• OpenFlow 1.3.1 is released and v1.4 in progress focuses on tunneling
and transport
• Ixia has joined ONF and is the Chair for Testing-Interop Working Group
24
OpenFlow
25
OpenFlow – Current Problem
26
OpenFlow - Solution
27
Carrier SDN with Segment Routing, BGP-LS and
PCEP
What Is Segment
Routing SDN
• Protocols
Innovation
• 24 RFC in two
years
• Unify DC + WAN
+ Aggregation
• Simplifies operation (lower
opex)
• Enables application-based
service creation (new revenue)
• Allows for better utilization of the
installed infrastructure (lower
capex)
29
Segment Routing Overview
• Forwarding state (segment) is established by IGP (ISIS or OSPF)
•
•
No LDP or RSVP-TE
Agnostic to forwarding data plane: IPv6 or MPLS
• MPLS Data plane is leveraged without any modification
•
•
push, swap and pop: all what we need
segment = label
• Source Routing
•
Source computes the path and then encodes path as a label or stack
of segments
• Architecturally designed to be integrated with application
• Provide optimum scalability, resiliency, simplicity and
virtualization
Benefit of SR over Classic MPLS
Simple
• Less numbers of protocols to
operate & troubleshoot (NO
LDP or RSVP)
• Less numbers of protocol
interactions to deal with
• Deliver automated FRR for
any topology
Ability to Scale
• Avoid millions of labels in
LDP database
• Avoid millions of RSVP TE
LSP’s in the network
• Avoid millions of tunnels to
configure
Segment Routing is SDN Ready
Smooth Integration with Controller
• SDN, PCE, Openflow
Perfect integration with application
• Simple and highly programmable
• Efficiency with guaranteed SLA
The state is no longer in the network
but in the packet
Segment Routing is SDN Solution for Carrier Network
Segment Routing – Technology Basic
• Simple extension to IGP (ISIS or OSPF), automatically builds and
maintain segments
•
•
Nodal Segment – A shortest path to the related node
Adjacency Segment – One hop through the related adjacency
• Excellent Scale: a node installs N+A FIB entries
• N = Nodal segment; A = Adjacency segment
Nodal Segment
•
•
Z advertise its node segment (loopback 0) ID 65 to all nodes in the IGP
domain
•
Simple ISIS sub-TLV extension
•
Assuming the same SRGB at every node (no label value shift from node to node)
All nodes install in their FIB the node segment ID 65 to Z in MPLS
dataplane
•
Node A Push 65 label in traffic
•
Node B, C also know to use 65 label to reach Z so they do Swap
•
Node D knows it’s the last hop and performs Pop
Adjacency Segment
• C allocates a local segment 9003 for CO link, means to “detour the path
to follow the interface CO”
• C advertises the adjacency segment in ISIS
• Simple sub-TLV extension for ISIS
• C is the only node to install the adjacency segment in MPLS dataplane
(FIB)
• Used for traffic engineering to steer traffic via a specific link
Segment Routing SDN
• MPLS or IPv6 data plane
• No RSVP-TE & LDP
• Centralized control plane
WAN Controller
BGP-LS RR
PCE
Other SR Domain
PCEP
BGP-LS speaker
PCC
OSPF/ISIS Segment Routing
Segment Routing use-case and targeted solutions
Simplify MPLS
WAN orchestration and optimization
Service Chaining SR
SDN WAN
controller
WAN
Automated 50msec protection
10
10
10
Application Enabled Routing
Peering orchestration
10
SDN Peering
controller
100
ISP
1
0
Carrier Ethernet simplification
BGP
T-LDP
BGP & SDN
RFC 3107
Autonomic &
Segment
Routing
RSVP-TE
MPLS-LDP
IGP & IP
MAKE IT SIMPLER & BETTER
© 2014 Cisco and/or its affiliates. All rights reserved.
ISP
Peer’s
DC
WAN
HOST
OPTIMIZING COST
EDCS-1463872.
ISP
NEW REVENUE
Cisco Confidential
37
NFV
NFV (Network Functions Virtualization)
NFV aims to deploy network functions on industry
standard high volume servers, switches and
storage
ETSI Industry Specification Group (ISG) for NFV
Created in 2012
• Has over 80 Active members and 120
participants
Benefits of NFV
•
•
•
•
•
Reduced CAPEX/OPEX
Flexibility and elasticity
Reduced time-to-market
Improved operational efficiency
Software-oriented innovation
Network
Functions
Virtualization
• SW flexibility
increases agility
• Concerns on
performance,
scale and
reliability
40
Validating NFV and Server Platform
NFV – New concepts and challenges
Elasticity &
Service Chaining
High-Level NFV Framework
VM Manager
vSwitch
Hypervisor
COTS HW
NICs
Drivers
NFV Performance Bottlenecks
Testing NFV using Hardware Test Ports
Testing L2/7 VNFs which are mapped to
TestingNIC
L2/3
Networkhas
Functions
physical
interfaces
many of the same
attributes of traditional testing:
• Forwarding performance (loss, latency,
throughput)
• Protocol performance & scale
• Multi-protocol/multi-dimensional testing
Unique testing and variables include:
• Performance of the vSwitch and VNF
• Determining the optimal resources
(CPU/memory) allocated to the virtual
appliance to meet the performance
requirement
• Instantiation of a service – how fast
• Termination of a service
• Reliability of a service
• Service isolation (affected by other VMs or
services on the server)
Testing NFV using Virtual Test Ports
Testing virtually by inserting test
Testing with Virtual Test Appliances
interfaces into the virtualized
server:
• Test the vSwitch for
performance
• Test each of the virtual
appliances
• Test virtual appliance chaining
• Isolate and test each function
before mapping to physical
interfaces
Can also test with a combination
of physical and virtual test ports
45
Questions