Infinity Hardware Resource Allocation Guide
Infinity Hardware Resource Allocation Guide
Infinity Hardware Resource Allocation Guide
Software Version 33
October 2023
Pexip Infinity Hardware Resource Allocation Guide
Contents
Introduction 3
Introduction
This guide describes how hardware resources are allocated in a Pexip Infinity deployment. It includes the following sections:
l Hardware resource allocation rules: how hardware resources are allocated and consumed in a Pexip Infinity deployment.
l Resource allocation examples: how resources are used in a number of specific scenarios, such as distributed and non-distributed
conferences, gateway calls, calls using standards-based endpoints, Skype for Business / Lync clients, or a combination of clients.
If you want to limit video calls to specific resolutions (and limit the transcoding node resources that are reserved for calls), you should
use the Maximum call quality setting.
The following rules determine how hardware resources are allocated and consumed by conference and gateway calls in a Pexip Infinity
deployment. They are measured in terms of HD resources, and assume that the Maximum call quality of the conference is set to HD
(more or less resources are used for Full HD and SD respectively, as described above):
l Standards-based endpoints (SIP or H.323) participants:
o Each participant uses 1 HD resource.
o Extra resources are consumed when sending main video and presentation content to the endpoint, than when compared to
only sending main video. The amount of additional resource depends on many factors, including the resolution and frame rate
of both sets of content, and the available bandwidth (plus the general maximum settings for call quality, bandwidth and
presentation bandwidth ratio). Thus, if for example, a Conferencing Node is using 1 HD of resource to send just main video,
and it then starts also sending presentation content, it may typically use a total of 1 HD of resource for presentation plus 0.5
HD for main video. However, in other circumstances (in particular, the bandwidth configured for the call), it could use 1 HD for
content plus 1 HD for video, or potentially just 0.5 HD for content and 0.5 HD for video.
o When sending two video streams to a dual-screen endpoint, the call bandwidth is always split 50-50 between video and
presentation content (and cannot be changed), and it uses twice the resource as when only sending main video.
o No additional resource is consumed when receiving presentation content from the endpoint.
l WebRTC VP8/VP9 participants:
o VP8 participants use 1 HD resource and VP9 participants use 1.25 HD resources for main video.
o Extra resources are consumed when sending main video and presentation content to the WebRTC endpoint, in the same
manner as described above for standards-based endpoints.
o No additional resource is consumed when receiving presentation content from the endpoint.
l Skype for Business / Lync participants:
o Each participant uses 1 HD resource.
o They each require 1 additional HD resource when sending or receiving presentation.
l If an endpoint is restricted due to bandwidth limitations to a lower resolution than the conference's maximum call quality, the
transcoding node will use the appropriate lower level of resource for that endpoint's connection. For example, if it is limited to an
SD connection it will use 0.5 HD resources.
l Each conference instance on each Transcoding Conferencing Node reserves a backplane connection at a resource level
corresponding to the conference's Maximum call quality setting, to allow the conference to become geographically distributed if
required. The exceptions to this are:
o Deployments with a single Conferencing Node. In such cases, no backplanes will ever be required, so capacity is not reserved.
o Conferences that are audio-only (in other words, where the conference has its Conference capabilities set to Audio-only). In
such cases, capacity equivalent to one audio connection is reserved for the backplane.
l Only one backplane connection is used for each conference on each Transcoding Conferencing Node, regardless of the number of
other transcoding nodes that are involved in the conference. Note that on the Administrator interface, backplanes are displayed as
unidirectional, so a single link between two nodes is shown with two backplanes, one in each direction. Likewise, a conference
between three nodes is shown with six backplanes. However, for resource allocation purposes each node would require just one
backplane connection.
l Pexip Infinity always tries to optimize gateway calls:
o A gateway call does not reserve resource for a backplane, but will use one if required (for example, if the participants are
connected via different Transcoding Conferencing Nodes).
o For a gateway call to Google Meet, the connection to Google Meet always uses 1 HD resource (it uses VP8) for main video.
The resources required for the VTC leg of the connection depend upon the type of endpoint and the Maximum call quality
setting. If the VTC endpoint starts to present content then an extra 1 HD resource is used for the connection from Pexip
Infinity to Google Meet. However, no additional resources are required on the Google Meet leg if presentation content is sent
from Google Meet, but 0.5 HD of additional resource would typically be required for each endpoint receiving presentation.
o For a gateway call to a Microsoft Teams meeting, the connection to Teams uses 1.5 HD of resource if Maximum call quality is
SD or HD, otherwise it uses 1.5 Full HD resources. The resources required for the VTC leg of the connection depend upon the
Maximum call quality setting. If any participant presents content, additional resources (typically 0.5 HD) would be required,
either on the Teams backplane (when an endpoint presents) or on the node handling the endpoint's media connection (when
a Teams client presents). The exact amount of resource used depends on the codec, resolution and frame rate of the
presentation stream.
l If an API participant is the first participant to join a conference, it will reserve a backplane for the conference.
In each of these cases, the hardware capacity requirements are described in terms of the HD resources required for a connection. A
connection can be a call or presentation from an endpoint to a Virtual Meeting Room or Virtual Auditorium, a backplane between
Transcoding Conferencing Nodes, or a call into or out of the Infinity Gateway. In this context, a connection is analogous to a port. Note
that a Skype for Business client may require two connections (one for the video call, and one for presentation content).
All of these examples are based around using the default Maximum call quality of HD. A Full HD connection typically uses twice the
resource of a HD connection and a standard definition SD connection typically uses half the resource. WebRTC connections are
assumed to be using the VP9 codec. VP9 calls consume around 1.25 times the resource for calls for main video (i.e. 1.25 HD resources
for 720p, and 2.5 HD for 1080p).
All of the examples assume that the endpoints are connecting directly to Transcoding Conferencing Nodes. However, the Teams
examples also show the effect of routing those calls via a Proxying Edge Node to the Teams Connector. (When a connection is proxied
via a Proxying Edge Node, the proxying node also consumes connection resources in order to forward the media streams on to a
Transcoding Conferencing Node. A transcoding node always consumes the same amount of connection resources regardless of
whether it has a direct connection to an endpoint, or it is receiving the media streams via a proxying node.)
l There are multiple Conferencing Nodes in a single system location, but the
conference is hosted on a single node.
l There are three endpoints (two standards-based and one WebRTC VP9*)
connecting directly to the node.
l Each endpoint uses HD video and audio.
Pexip Infinity requires 4.25 HD resources. 3.25 HD resources are required by the
endpoints (the two standards-based endpoints each use 1 HD resource and the
WebRTC VP9 client uses 1.25 HD), and 1 HD resource is reserved by the backplane.
l A presentation is then sent from a standards-based endpoint (see diagram).
Pexip Infinity then typically uses an extra 0.5 HD resource for each standards-based or
WebRTC endpoint that is receiving the presentation, meaning the total resources used
on Node 1 is now 5.25 HD.
The conference uses 3 concurrent call licenses.
Pexip Infinity uses an extra 0.5 HD resource on Node 1 for the WebRTC
client to receive presentation, meaning the total resources used on Node
1 is now 3.75 HD. And it requires 1 additional HD resource on Node 2 for the SfB client to receive presentation, so the total resources
used on Node 2 is now 3 HD.
The conference uses 3 concurrent call licenses.
Pexip Infinity requires 2.5 HD resources. The standards-based endpoint uses 1 HD resource, and
the backplane connection to the Teams Connector uses 1.5 HD resources.
l A presentation is then sent from the standards-based endpoint (see diagram).
Pexip Infinity would typically use an extra 0.5 HD resource on the Teams backplane connection for the Microsoft Teams meeting to
receive presentation. If one of the Teams clients sent presentation content it would instead typically use an extra 0.5 HD resource for
the standards-based endpoint to receive presentation.
The call uses 2 concurrent call licenses in both cases.
Single gateway call via Proxying Edge Node (standards-based endpoint to Teams)
Pexip Infinity requires 0.2 HD resources on the proxying node, and 2.5
HD resources on the transcoding node (1 for the standards-based
endpoint and 1.5 for the backplane connection).
l A presentation is then sent from the standards-based endpoint (see diagram).
Pexip Infinity would typically use an extra 0.5 HD resource on the Teams backplane connection for the Microsoft Teams meeting to
receive presentation. If one of the Teams clients sent presentation content it would instead typically use an extra 0.5 HD resource for
the standards-based endpoint to receive presentation.
The call uses 2 concurrent call licenses in both cases.
Multiple distributed gateway calls via Proxying Edge Node (mixed endpoints to same Teams
meeting)
l There is a proxying node in location A and a transcoding node in location B.
l Two gateway calls are placed from endpoints connected to the transcoding node to a Microsoft Teams meeting. One endpoint is a
standards-based endpoint and the other is a Skype for Business client. Both calls are routed via the proxying node and the Teams
Connector.
l Two Teams clients are also connected to the Microsoft Teams meeting.
l Each endpoint uses HD video and audio.
Pexip Infinity requires 0.4 HD resources on the proxying node and 5 HD resources on the transcoding node (the standards-based
endpoint requires 1 for its connection to the gateway, and 1.5 for the backplane; the Skype for Business client requires 1 for its
connection to the gateway, and 1.5 for the backplane) (see diagram above).
l A presentation is then sent from the standards-based endpoint (see diagram below).
Pexip Infinity would typically use an extra 0.5 HD resource on the Teams backplane connection for the Microsoft Teams meeting to
receive presentation. The Skype for Business client requires 1 additional HD resource on the transcoding node to receive presentation.
* WebRTC connections are assumed to be using the VP9 codec. VP9 calls consume around 1.25 times the resource for calls for main
video (i.e. 1.25 HD resources for 720p, and 2.5 HD for 1080p).
Pexip Infinity requires 2 HD resources. The standards-based endpoint uses 1 HD resource. The
connection to Google Meet uses 1 HD resource (it uses VP8).
l A presentation is then sent from the standards-based endpoint (see diagram).
Pexip Infinity requires an additional 1 HD resource for Google Meet to receive presentation.
However, no additional resources are required on the Google Meet leg if presentation content
is sent from Google Meet, but 0.5 HD of additional resource would typically be required for
each endpoint receiving presentation.
The call uses 2 concurrent call licenses in both cases.
Pexip Infinity requires an additional 1 HD resource for Google Meet to receive presentation. However, no additional resources are
required on the Google Meet leg if presentation content is sent from Google Meet, but 0.5 HD of additional resource would typically
be required for each endpoint receiving presentation.
The call uses 2 concurrent call licenses in both cases.