[jitsi-users] Best EC2 Instance for Jitsi Meet


#1

Hi guys,

Trying to get the best out of Jitsi meet for our custom installation on AWS EC2 and was wondering if anyone did experiments on the tradeoffs regarding the different machine types. Given that on our use cases most connections will be P2P (conference between 2 users) it should not be an issue. However, my tests with t2.micro showed that even with 1 single call at the same time, the video freezes quite often (when not P2P).

I think CPU and network throughput are the most important factors. But given a scenario where I need to support around 10 simultaneous calls (let’s even consider they are not P2P) on an average resolution of 640x320, which machine would you recommend on EC2?

I know I could test this on my own but I assume someone has had this challenge before, so I opted for asking first.

Best,
Jose


#2

Hi Jose,

Hi guys,

Trying to get the best out of Jitsi meet for our custom installation on AWS EC2 and was wondering if anyone did experiments on the tradeoffs regarding the different machine types. Given that on our use cases most connections will be P2P (conference between 2 users) it should not be an issue. However, my tests with t2.micro showed that even with 1 single call at the same time, the video freezes quite often (when not P2P).

I think CPU and network throughput are the most important factors. But given a scenario where I need to support around 10 simultaneous calls (let’s even consider they are not P2P) on an average resolution of 640x320, which machine would you recommend on EC2?

I know I could test this on my own but I assume someone has had this challenge before, so I opted for asking first.

I don't have any useful data to share, but I expect that 10 simultaneous call will not require a big machine (unless they have many participants). So I would start testing with a small machine, monitoring its CPU load and the loss reported by clients.

Note that freezes are not necessarily caused by the VM being too small.

Regards,
Boris

···

On 23/06/2017 14:04, Jose Simoes wrote:


#3

Hi Boris,

Thanks for the prompt reply. I would expect the same, however, on the same conditions (device / connection) when using P2P the quality is ok without interruptions and when not (on the t2.micro - the free tier one), the video stops for a few seconds sometimes (showing that grey message). Monitoring CPU, even for that instance, it only consumes 10%, so I would say that network should be the issue.

That is why I asked if there is a minimum threshold so I could skip testing :slight_smile: If anyone else has experience on AWS EC2, feel free to comment.

Best,
Jose

···

On 23 Jun 2017, at 16:18, Boris Grozev <boris@jitsi.org> wrote:

Hi Jose,

On 23/06/2017 14:04, Jose Simoes wrote:

Hi guys,
Trying to get the best out of Jitsi meet for our custom installation on AWS EC2 and was wondering if anyone did experiments on the tradeoffs regarding the different machine types. Given that on our use cases most connections will be P2P (conference between 2 users) it should not be an issue. However, my tests with t2.micro showed that even with 1 single call at the same time, the video freezes quite often (when not P2P).
I think CPU and network throughput are the most important factors. But given a scenario where I need to support around 10 simultaneous calls (let’s even consider they are not P2P) on an average resolution of 640x320, which machine would you recommend on EC2?
I know I could test this on my own but I assume someone has had this challenge before, so I opted for asking first.

I don't have any useful data to share, but I expect that 10 simultaneous call will not require a big machine (unless they have many participants). So I would start testing with a small machine, monitoring its CPU load and the loss reported by clients.

Note that freezes are not necessarily caused by the VM being too small.

Regards,
Boris


#4

I also wanted to run it on a cloud environment, in my case DO, but finally
decided to go bare metal cloud, precisely to avoid virtualization
limitations.
I am amazed that all the Jitsi meet components would even run on such a
minimalistic instance, may I ask why do you use ECS?
Do you MUST use ec2 or can you look for more options?
In any case I wouldn't use anything smaller than a t2.small, not even for
testing!

···

Best, Yosu Cadilla ツ

Consultor de màrqueting @ Collita Digital - http://www.collitadigital.cat
CMO @ low-cost.marketing - http://low-cost.marketing/
Content Engineer @ Content.engineering - http://content.engineering/
Facilitator for CloudCamp in Europe and LATAM.

Phone: (+34) 693 481 365
Skype: new_yosu
Twitter: @_YC

On 23 June 2017 at 21:22, Jose Simoes <simoes83@gmail.com> wrote:

Hi Boris,

Thanks for the prompt reply. I would expect the same, however, on the same
conditions (device / connection) when using P2P the quality is ok without
interruptions and when not (on the t2.micro - the free tier one), the video
stops for a few seconds sometimes (showing that grey message). Monitoring
CPU, even for that instance, it only consumes 10%, so I would say that
network should be the issue.

That is why I asked if there is a minimum threshold so I could skip
testing :slight_smile: If anyone else has experience on AWS EC2, feel free to comment.

Best,
Jose

> On 23 Jun 2017, at 16:18, Boris Grozev <boris@jitsi.org> wrote:
>
> Hi Jose,
>
> On 23/06/2017 14:04, Jose Simoes wrote:
>> Hi guys,
>> Trying to get the best out of Jitsi meet for our custom installation on
AWS EC2 and was wondering if anyone did experiments on the tradeoffs
regarding the different machine types. Given that on our use cases most
connections will be P2P (conference between 2 users) it should not be an
issue. However, my tests with t2.micro showed that even with 1 single call
at the same time, the video freezes quite often (when not P2P).
>> I think CPU and network throughput are the most important factors. But
given a scenario where I need to support around 10 simultaneous calls
(let’s even consider they are not P2P) on an average resolution of 640x320,
which machine would you recommend on EC2?
>> I know I could test this on my own but I assume someone has had this
challenge before, so I opted for asking first.
>
> I don't have any useful data to share, but I expect that 10 simultaneous
call will not require a big machine (unless they have many participants).
So I would start testing with a small machine, monitoring its CPU load and
the loss reported by clients.
>
> Note that freezes are not necessarily caused by the VM being too small.
>
> Regards,
> Boris

_______________________________________________
users mailing list
users@jitsi.org
Unsubscribe instructions and other list options:
http://lists.jitsi.org/mailman/listinfo/users


#5

Not necessarily on AWS EC2. We are building a telehealth service that needs to be on top of a scalable architecture (bare metal is not the best option). Our current setup is on Azure but we are now testing Jitsi as an alternative for the long term. I guess the team has more experience managing AWS and it helps that as a startup we are entitled to some free credits as well :slight_smile:

Best,
Jose

···

On 23 Jun 2017, at 16:42, Yosu Cadilla ツ <yosu.cadilla@gmail.com> wrote:

I also wanted to run it on a cloud environment, in my case DO, but finally decided to go bare metal cloud, precisely to avoid virtualization limitations.
I am amazed that all the Jitsi meet components would even run on such a minimalistic instance, may I ask why do you use ECS?
Do you MUST use ec2 or can you look for more options?
In any case I wouldn't use anything smaller than a t2.small, not even for testing!

Best, Yosu Cadilla ツ

Consultor de màrqueting @ Collita Digital - http://www.collitadigital.cat <http://www.collitadigital.cat/>
CMO @ low-cost.marketing - http://low-cost.marketing/
Content Engineer @ Content.engineering - http://content.engineering/
Facilitator for CloudCamp in Europe and LATAM.

Phone: (+34) 693 481 365
Skype: new_yosu
Twitter: @_YC

On 23 June 2017 at 21:22, Jose Simoes <simoes83@gmail.com <mailto:simoes83@gmail.com>> wrote:
Hi Boris,

Thanks for the prompt reply. I would expect the same, however, on the same conditions (device / connection) when using P2P the quality is ok without interruptions and when not (on the t2.micro - the free tier one), the video stops for a few seconds sometimes (showing that grey message). Monitoring CPU, even for that instance, it only consumes 10%, so I would say that network should be the issue.

That is why I asked if there is a minimum threshold so I could skip testing :slight_smile: If anyone else has experience on AWS EC2, feel free to comment.

Best,
Jose

> On 23 Jun 2017, at 16:18, Boris Grozev <boris@jitsi.org <mailto:boris@jitsi.org>> wrote:
>
> Hi Jose,
>
> On 23/06/2017 14:04, Jose Simoes wrote:
>> Hi guys,
>> Trying to get the best out of Jitsi meet for our custom installation on AWS EC2 and was wondering if anyone did experiments on the tradeoffs regarding the different machine types. Given that on our use cases most connections will be P2P (conference between 2 users) it should not be an issue. However, my tests with t2.micro showed that even with 1 single call at the same time, the video freezes quite often (when not P2P).
>> I think CPU and network throughput are the most important factors. But given a scenario where I need to support around 10 simultaneous calls (let’s even consider they are not P2P) on an average resolution of 640x320, which machine would you recommend on EC2?
>> I know I could test this on my own but I assume someone has had this challenge before, so I opted for asking first.
>
> I don't have any useful data to share, but I expect that 10 simultaneous call will not require a big machine (unless they have many participants). So I would start testing with a small machine, monitoring its CPU load and the loss reported by clients.
>
> Note that freezes are not necessarily caused by the VM being too small.
>
> Regards,
> Boris

_______________________________________________
users mailing list
users@jitsi.org <mailto:users@jitsi.org>
Unsubscribe instructions and other list options:
http://lists.jitsi.org/mailman/listinfo/users
_______________________________________________
users mailing list
users@jitsi.org
Unsubscribe instructions and other list options:
http://lists.jitsi.org/mailman/listinfo/users