Jump to: navigation, search

Difference between revisions of "TelcoWorkingGroup/UseCases"

(→‎Session Border Controller)
 
(44 intermediate revisions by 7 users not shown)
Line 1: Line 1:
  
−
=Overview =
+
=Contributing Use Cases=
 +
 
 +
The Telecommunications Working group welcomes use cases from Communication Service Providers (CSPs), Network Equipment Providers (NEPs) and other organizations in the telecommunications industry. To begin adding a use case simply copy the "Template" section of this page to the bottom of the list and rename it to a name that describes your use case.
 +
 
 +
When writing use cases focus on "what" you want to do and "why" rather than specific OpenStack requirements or solutions. Our aim as a working group is to assist in distilling those requirements or solutions from the use cases presented to ensure that we are building functionality that benefits all relevant telecommunications use cases. Submission of use cases that pertain to different implementations of the same network function (e.g. vEPC) are welcome as are use cases that speak to the more general demands telecommunications workloads place upon the infrastructure that supports them. In this initial phase of use case analysis the intent is to focus on those workloads that run on top of the provided infrastructure before moving focus to other areas.
 +
 
 +
Use cases are now written in [http://docutils.sourceforge.net/rst.html ReStructured Text] format and stored in the [http://git.openstack.org/cgit/stackforge/telcowg-usecases/ telcowg-usecases] git repository on Stackforge.
 +
 
 +
=Reviewing Use Cases=
 +
 
 +
The working group uses [http://review.openstack.org OpenStack's Gerrit installation] to collaborate on use case documentation, with the resultant work ultimately being stored [http://git.openstack.org/cgit/stackforge/telcowg-usecases/ in a git repository]. To review items stored in Gerrit you will first need to [[Gerrit_Workflow#Account_Setup|create an account]].
 +
 
 +
Note that to simply review items you will not need to sign the CLA, you will need to do this to upload use cases though. If you have any concerns about this process, consider joining one of the weekly [[TelcoWorkingGroup]] meetings to ask for assistance.
 +
 
 +
Once you have created an account you can find open items for review by opening this query in your web browser:
 +
 
 +
* https://review.openstack.org/#/q/status:open+project:stackforge/telcowg-usecases,n,z
 +
 
 +
The result of which will look something like this:
 +
 
 +
<gallery>
 +
Telcowg-user-cases-screen-1.png|Example telcowg-usecases query result.
 +
</gallery>
 +
 
 +
=Updating Use Cases=
  
 
=Contributed Use Cases=
 
=Contributed Use Cases=
Line 7: Line 31:
  
 
=== Description ===
 
=== Description ===
 +
 +
''Describe the use case in terms of what's being done and why.''
  
 
=== Characteristics ===
 
=== Characteristics ===
  
−
=== Requirements ===
+
''Describe important characteristics of the use case.''
 +
 
 +
==VPN Instantiation==
  
−
==Session Border Controller==
+
Contributed by Margaret Chiosi
  
−
Contributed by: Calum Loudon
+
Etherpad: https://etherpad.openstack.org/p/telcowg-usecase-VPN_Instantiation
  
 
===Description===
 
===Description===
  
−
Perimeta Session Border Controller, Metaswitch Networks. Sits on the edge of a service provider's network and polices SIP and RTP (i.e. VoIP) control and media traffic passing over the access network between end-users and the core network or the trunk network between the core and another SP.
+
VPN services are critical for the enterprise market which the Telcos provide services to. As we look to virtualize our PEs, VPN instantiation on a vPE needs to be addressed since connectivity is important. Proposal is to focus on ODL/Neutron linkage to openstack orchestration.
 +
Instantiate a VPN service on a vPE connecting to either a vPE or PE. This includes identifying where the vPE needs to be located (some set of criteria needs to be defined  - latency, diversity..) and then created on a virtualized environment. Connectivity to the other vPE/PEs need to be setup. Then finally the VPN service over the different vPE/PE which match the customer sites needs to get instantiated.
  
 
===Characteristics===
 
===Characteristics===
  
−
* Fast and guaranteed performance:
+
*Affinity rules
−
** Performance in the order of several million VoIP packets (~64-220 bytes depending on codec) per second per core (achievable on COTS hardware).
+
*ODL SDN Controller for connectivity setup
−
** Guarantees provided via SLAs.
+
*Physical connectivity between the different vPE/PE environments are assumed to exist
−
* Fully high availability
+
*Logical connectivity between different vPE/PE needs to be setup as the vPE is instantiated
−
** No single point of failure, service continuity over both software and hardware failures.
+
*VPN service connectivity needs to be setup
−
* Elastically scalable
+
*need to add the flow logic between the openstack components and ODL
−
** NFV orchestrator adds and removes instances in response to network demands.
 
−
* Traffic segregation (ideally)
 
−
** Separate traffic from different customers via VLANs.
 
  
 
===Requirements===
 
===Requirements===
  
−
* Fast & guaranteed performance (network)
+
*Affinity rules
−
** Packets per second target -> either SR-IOV or an accelerated DPDK-like data plane:
+
*ODL SDN Controller for connectivity setup
−
*** "SR-IOV Networking Support" (https://blueprints.launchpad.net/nova/+spec/pci-passthrough-sriov) -completed 2014.2
+
*Physical connectivity between the different vPE/PE environments are assumed to exist
−
*** "Open vSwitch to use patch ports" (https://blueprints.launchpad.net/neutron/+spec/openvswitch-patch-port-use) - patched 2014.2
+
*Logical connectivity between different vPE/PE needs to be setup as the vPE is instantiated
−
*** "userspace vhost in ovd vif bindings" (https://blueprints.launchpad.net/nova/+spec/libvirt-ovs-use-usvhost) - BP says Abandoned?
+
*VPN service connectivity needs to be setup
−
*** "Snabb NFV driver" (https://blueprints.launchpad.net/neutron/+spec/snabb-nfv-mech-driver)  - Not clear reasons of hold up (-2status )?
+
*Don't need to setup connectivity to customer router (CE) for this use case
−
*** "VIF_VHOSTUSER"  (https://blueprints.launchpad.net/nova/+spec/vif-vhostuser) -In code review, awaiting approval, kilo-1?
 
  
−
* Fast & guaranteed performance (compute):
+
==Session Border Controller==
−
** To optimize data rate we need to keep all working data in L3 cache:
 
−
***"Virt driver pinning guest vCPUs to host pCPUs" (https://blueprints.launchpad.net/nova/+spec/virt-driver-cpu-pinning) - Needs code review, kilo-1?
 
−
** To optimize data rate need to bind to NIC on host CPU's bus:
 
−
*** "I/O (PCIe) Based NUMA Scheduling" (https://blueprints.launchpad.net/nova/+spec/input-output-based-numa-scheduling) - targeted kilo-1
 
−
**To offer guaranteed performance as opposed to 'best efforts' we need:
 
−
** To control placement of cores, minimise TLB misses and get accurate info about core topology (threads vs. hyperthreads etc.); maps to the remaining blueprints on NUMA & vCPU topology:
 
−
*** "Virt driver guest vCPU topology configuration" (https://blueprints.launchpad.net/nova/+spec/virt-driver-vcpu-topology) - implemented 2014.2
 
−
*** "Virt driver guest NUMA node placement & topology" (https://blueprints.launchpad.net/nova/+spec/virt-driver-numa-placement) - implemented 2014.2
 
−
*** "Virt driver large page allocation for guest RAM" (https://blueprints.launchpad.net/nova/+spec/virt-driver-large-pages) - proposed kilo, target for kilo-1
 
−
** May need support to prevent 'noisy neighbours' stealing L3 cache - unproven, and no blueprint we're aware of.
 
  
−
* High availability:
+
Contributed by: Calum Loudon
−
** Requires anti-affinity rules to prevent active/passive being instantiated on same host - already supported, so no gap.
 
  
−
* Elastic scaling:
+
Review: https://review.openstack.org/#/c/176301/
−
** Readily achievable using existing features - no gap.
 
  
−
* VLAN trunking:
+
==Virtual IMS Core==
−
** "VLAN trunking networks for NFV" (https://blueprints.launchpad.net/neutron/+spec/nfv-vlan-trunks et al). - Needs resubmission by Ian Wells for approval
+
 
 +
Contributed by: Calum Loudon
 +
 
 +
Review: https://review.openstack.org/#/c/158997/
 +
 
 +
== Access to physical network resources ==
 +
 
 +
Contributed by: Jannis Rake-Revelant
  
−
* "GTP tunnel support for mobile network for NFV  VNFs like SGW, PGW, MME (https://blueprints.launchpad.net/neutron/+spec/provider-network.type-gtp -needs update for resubmission)
+
Etherpad: https://etherpad.openstack.org/p/telcowg-usecase-Access_to_physical_network
  
−
* Other:
+
=== Description ===
−
** Being able to offer apparent traffic separation (e.g. service traffic vs. application management) over single network is also useful in some cases.
 
−
*** "Support two interfaces from one VM attached to the same network" (https://blueprints.launchpad.net/nova/+spec/multiple-if-1-net) - implemented 2014.2
 
  
−
==Virtual IMS Core==
+
This use case aims to solve the problem of accessing physical (network) devices outside of the Openstack Infrastructure, that are '''not''' addressable by a public IP address. This use case can currently be implemented in various ways, as will be detailed later on.
 +
The background of this use case is the necessity to communicate with physical devices, in our case e.g. an eNodeB, to a VNF, e.g. a vEPC. Communication/ addressability should be possible from either side. In the current environment different physical devices are separated by VLANs and private IP subnets. The goal is to establish L3 (or L2 if that is "easier") connectivity.
  
−
Contributed by: Calum Loudon
 
  
−
===Description===
+
The main goal of this use case is not necessarily to implement something new but to discuss the practicability of the current implementations. If I missed an alternative implementation please add it to the list.
  
−
Project Clearwater, http://www.projectclearwater.org/.  An open source implementation of an IMS core designed to run in the cloud and be massively scalable.  It provides SIP-based call control for voice and video as well as SIP-based messaging apps.  As an IMS core it provides P/I/S-CSCF function together with a BGCF and an HSS cache, and includes
+
=== Characteristics ===
−
a WebRTC gateway providing interworking between WebRTC & SIP clients.
 
  
−
===Characteristics relevant to NFV/OpenStack===
+
Possible current implementations include:
  
−
* Mainly a compute application: modest demands on storage and networking.
+
* L3 gateways
−
* Fully HA, with no SPOFs and service continuity over software and hardware failures; must be able to offer SLAs.
+
** SNAT
−
* Elastically scalable by adding/removing instances under the control of the NFV orchestrator.
+
** L3 forwarding
 +
** Floating IPs
 +
* External provider networks, e.g. VLAN backed
 +
* L2 gateways, currently only possible with 3rd party software (?)
  
−
===Requirements===
+
=== References ===
 +
* [http://docbox.etsi.org/ISG/NFV/Open/Latest_Drafts/NFV-PER001v009%20-%20NFV%20Performance%20&%20Portability%20Best%20Practises.pdf Network Functions Virtualization NFV Performance & Portability Best Practices - DRAFT]
 +
* ETSI-NFV Use Cases V1.1.1 [http://www.etsi.org/deliver/etsi_gs/NFV/001_099/001/01.01.01_60/gs_NFV001v010101p.pdf]
  
−
* Compute application:
+
==Security Segregation (Placement Zones)==
−
** OpenStack already provides everything needed; in particular, there are no requirements for an accelerated data plane, nor for core pinning nor NUMA
 
  
−
*HA:
+
Contributed by: Daniel Schabarum (DaSchab)
−
** implemented as a series of N+k compute pools; meeting a given SLA requires being able to limit the impact of a single host failure
 
−
** potentially a scheduler gap here: affinity/anti-affinity can be expressed pair-wise between VMs, which is sufficient for a 1:1 active/passive architecture, but an N+k pool needs a concept equivalent to "group anti-affinity" i.e. allowing the NFV orchestrator to assign each VM in a pool to one of X buckets, and requesting OpenStack to ensure no single host failure can affect more than one bucket
 
−
** (there are other approaches which achieve the same end e.g. defining a group where the scheduler ensures every pair of VMs within that group are not instantiated on the same host)
 
−
** for study whether this can be implemented using current scheduler hints
 
  
−
* Elastic scaling:
+
Review: https://review.openstack.org/#/c/163399
−
** as for compute requirements there is no gap - OpenStack already provides everything needed.
 
  
−
== VLAN Trunking ==
+
= Work In Progress =
  
 +
== Service Chaining ==
  
−
The big picture is that this is about how service providers can use
+
Etherpad: https://etherpad.openstack.org/p/kKIqu2ipN6
−
virtualisation to provide differentiated network services to their customers
 
−
(and specifically enterprise customers rather than end users); it's not about
 
−
VMs want to set up networking between themselves.
 
  
−
A typical service provider may be providing network services to thousands or
+
== Orchestration ==
−
more of enterprise customers.  The details of and configuration required for
 
−
individual services will differ from customer to customer.  For example,
 
−
consider a Session Border Control service (basically, policing VoIP
 
−
interconnect): different customers will have different sets of SIP trunks that
 
−
they can connect to, different traffic shaping requirements, different
 
−
transcoding rules etc.
 
  
−
Those customers will normally connect in to the service provider in one of two
+
Etherpad: https://etherpad.openstack.org/p/telco_orchestration
−
ways: a dedicated physical link, or through a VPN over the public Internet. 
 
−
Once that traffic reaches the edge of the SP's network, then it makes sense for
 
−
the SP to put all that traffic onto the same core network while keeping some
 
−
form of separation to allow the network services to identify the source of the
 
−
traffic and treat it independently.  There are various overlay techniques that
 
−
can be used (e.g. VXLAN, GRE tunnelling) but one common and simple one is VLANs.
 
−
Carrying VLAN trunking into the VM allows this scheme to continue to be used in
 
−
a virtual world.
 
  
−
In this set-up, then any VMs implementing those services have to be able to
+
== MNO/MVNO Use Case==
−
differentiate between customers.  About the only way of doing that today in
 
−
OpenStack is to configure one provider network per customer then have one vNIC
 
−
per provider network, but that approach clearly doesn't scale (both performance
 
−
and configuration effort) if a VM has to see traffic from hundreds or thousands
 
−
of customers.  Instead, carrying VLAN trunking into the VM allows them to do
 
−
this scalably.
 
  
−
The net is that a VM providing a service that needs to have access to a
+
Etherpad: https://etherpad.openstack.org/p/mno-mvno
−
customer's non-NATed source addresses needs an overlay technology to allow this,
 
−
and VLAN trunking into the VM is sufficiently scalable for this use case and
 
−
leverages a common approach.
 
  
−
From: http://lists.openstack.org/pipermail/openstack-dev/2014-October/047548.html
+
== SIP Load-Balancing-as-a-Service==
  
−
= References: =
+
Etherpad: https://etherpad.openstack.org/p/telcowg-usecase-SIP_LBaaS
−
* [http://docbox.etsi.org/ISG/NFV/Open/Latest_Drafts/NFV-PER001v009%20-%20NFV%20Performance%20&%20Portability%20Best%20Practises.pdf Network Functions Virtualization NFV Performance & Portability Best Practices - DRAFT]
 
−
* ETSI-NFV Use Cases V1.1.1 [http://www.etsi.org/deliver/etsi_gs/NFV/001_099/001/01.01.01_60/gs_NFV001v010101p.pdf]
 

Latest revision as of 13:27, 22 April 2015

Contributing Use Cases

The Telecommunications Working group welcomes use cases from Communication Service Providers (CSPs), Network Equipment Providers (NEPs) and other organizations in the telecommunications industry. To begin adding a use case simply copy the "Template" section of this page to the bottom of the list and rename it to a name that describes your use case.

When writing use cases focus on "what" you want to do and "why" rather than specific OpenStack requirements or solutions. Our aim as a working group is to assist in distilling those requirements or solutions from the use cases presented to ensure that we are building functionality that benefits all relevant telecommunications use cases. Submission of use cases that pertain to different implementations of the same network function (e.g. vEPC) are welcome as are use cases that speak to the more general demands telecommunications workloads place upon the infrastructure that supports them. In this initial phase of use case analysis the intent is to focus on those workloads that run on top of the provided infrastructure before moving focus to other areas.

Use cases are now written in ReStructured Text format and stored in the telcowg-usecases git repository on Stackforge.

Reviewing Use Cases

The working group uses OpenStack's Gerrit installation to collaborate on use case documentation, with the resultant work ultimately being stored in a git repository. To review items stored in Gerrit you will first need to create an account.

Note that to simply review items you will not need to sign the CLA, you will need to do this to upload use cases though. If you have any concerns about this process, consider joining one of the weekly TelcoWorkingGroup meetings to ask for assistance.

Once you have created an account you can find open items for review by opening this query in your web browser:

The result of which will look something like this:

  • Example telcowg-usecases query result.

Updating Use Cases

Contributed Use Cases

Template

Description

Describe the use case in terms of what's being done and why.

Characteristics

Describe important characteristics of the use case.

VPN Instantiation

Contributed by Margaret Chiosi

Etherpad: https://etherpad.openstack.org/p/telcowg-usecase-VPN_Instantiation

Description

VPN services are critical for the enterprise market which the Telcos provide services to. As we look to virtualize our PEs, VPN instantiation on a vPE needs to be addressed since connectivity is important. Proposal is to focus on ODL/Neutron linkage to openstack orchestration. Instantiate a VPN service on a vPE connecting to either a vPE or PE. This includes identifying where the vPE needs to be located (some set of criteria needs to be defined - latency, diversity..) and then created on a virtualized environment. Connectivity to the other vPE/PEs need to be setup. Then finally the VPN service over the different vPE/PE which match the customer sites needs to get instantiated.

Characteristics

  • Affinity rules
  • ODL SDN Controller for connectivity setup
  • Physical connectivity between the different vPE/PE environments are assumed to exist
  • Logical connectivity between different vPE/PE needs to be setup as the vPE is instantiated
  • VPN service connectivity needs to be setup
  • need to add the flow logic between the openstack components and ODL

Requirements

  • Affinity rules
  • ODL SDN Controller for connectivity setup
  • Physical connectivity between the different vPE/PE environments are assumed to exist
  • Logical connectivity between different vPE/PE needs to be setup as the vPE is instantiated
  • VPN service connectivity needs to be setup
  • Don't need to setup connectivity to customer router (CE) for this use case

Session Border Controller

Contributed by: Calum Loudon

Review: https://review.openstack.org/#/c/176301/

Virtual IMS Core

Contributed by: Calum Loudon

Review: https://review.openstack.org/#/c/158997/

Access to physical network resources

Contributed by: Jannis Rake-Revelant

Etherpad: https://etherpad.openstack.org/p/telcowg-usecase-Access_to_physical_network

Description

This use case aims to solve the problem of accessing physical (network) devices outside of the Openstack Infrastructure, that are not addressable by a public IP address. This use case can currently be implemented in various ways, as will be detailed later on. The background of this use case is the necessity to communicate with physical devices, in our case e.g. an eNodeB, to a VNF, e.g. a vEPC. Communication/ addressability should be possible from either side. In the current environment different physical devices are separated by VLANs and private IP subnets. The goal is to establish L3 (or L2 if that is "easier") connectivity.


The main goal of this use case is not necessarily to implement something new but to discuss the practicability of the current implementations. If I missed an alternative implementation please add it to the list.

Characteristics

Possible current implementations include:

  • L3 gateways
    • SNAT
    • L3 forwarding
    • Floating IPs
  • External provider networks, e.g. VLAN backed
  • L2 gateways, currently only possible with 3rd party software (?)

References

Security Segregation (Placement Zones)

Contributed by: Daniel Schabarum (DaSchab)

Review: https://review.openstack.org/#/c/163399

Work In Progress

Service Chaining

Etherpad: https://etherpad.openstack.org/p/kKIqu2ipN6

Orchestration

Etherpad: https://etherpad.openstack.org/p/telco_orchestration

MNO/MVNO Use Case

Etherpad: https://etherpad.openstack.org/p/mno-mvno

SIP Load-Balancing-as-a-Service

Etherpad: https://etherpad.openstack.org/p/telcowg-usecase-SIP_LBaaS

Retrieved from "https://wiki.openstack.org/w/index.php?title=TelcoWorkingGroup/UseCases&oldid=78013"