Hello,
I am running 3 Large Instance EC2 servers with Origin/Edge setup (1 origin, 2 edges) and load balancing solution. I was wondering how many concurrent streams a large ec2 server can handle. And how many concurrent users per stream?
Also, what sort of information will the Wowza load tester tool give?
I was wondering how many concurrent streams a large ec2 server can handle. And how many concurrent users per stream?
These are really the same question, unless the first question refers to incoming streams? If so, you will have to test, but probably not many on a m1.large, 10 - 20 at the most is my guess, but I am not sure.
If you just mean total streams, the m1.large gets about 250kbs bandwidth throughput. So if you have a 500kbs live stream, or your files are about 500kbs, you can stream to about 500 clients concurrently.
Richard
Again, I do not know. 10 - 20 is a guess. It might be lower. I doubt it is higher, but maybe. You have to test.
Richard
You can have many subscribers to one stream, which Wowza is essentially repeating to each unicast subscriber. The usual limiting factor is bandwidth. But the number of unique incoming streams has other processor limits.
Richard
There isn’t anything to go by for this, or testing tool. I can tell you that getting 100 unique incoming streams is a lot. I have heard of higher numbers, but not much higher. Of course you need a lot of ram and a late model dual quad cpu is recommended… 64bit OS/Java, well tuned
Richard
Incoming streams that are being transcoded will be part of your incoming stream count. Transcoding will require additional CPU. Take a look at this Trasncoder AddOn Performance Benchmark article for guidance on what kind of transcoding performance you can expect with regard to transcoding. Again, you should test with your streams in your setup to determine what performance you can expect.
-Lisa
These are really the same question, unless the first question refers to incoming streams? If so, you will have to test, but probably not many on a m1.large, 10 - 20 at the most is my guess, but I am not sure.
If you just mean total streams, the m1.large gets about 250kbs bandwidth throughput. So if you have a 500kbs live stream, or your files are about 500kbs, you can stream to about 500 clients concurrently.
Richard
Yeah, I was referring how many live streams I can encode to the origin server at the same time. So I would be able to get up to 10-20 live streams encoding to the origin server?
And if I want to increase this, I would just need to add additional origins correct? And Not edges since that is for the clients to connect to?
Thanks.
Thank you for your reply Richard. We are also transcoding our incoming streams as well using the wowza transcoder on our origin. Would you say that will impact the numbers of incoming streams or this wouldn’t be a factor?
Thanks.
Thank you for your reply Richard. We are also transcoding our incoming streams as well using the wowza transcoder on our origin. Would you say that will impact the numbers of incoming streams or this wouldn’t be a factor?
Thanks.
Hi Richard,
Can you please reply with whether transcoding may have an impact on the origin for incoming streams?
Thanks.
These are really the same question, unless the first question refers to incoming streams? If so, you will have to test, but probably not many on a m1.large, 10 - 20 at the most is my guess, but I am not sure.
If you just mean total streams, the m1.large gets about 250kbs bandwidth throughput. So if you have a 500kbs live stream, or your files are about 500kbs, you can stream to about 500 clients concurrently.
Richard
Hi, sorry to hijack this thread, but what are the factors that mean there is such a disparity between expected outbound vs. inbound streams?
Is there any more information available as to how this can be estimated?
I’ve looked at several threads on this forum, but there doesn’t seem to be much concrete advice as to what spec server can handle in terms of incoming streams.