we have a serious problem with the load average cpu server.
We have had a EC2 Amazon server model r3.large and 60 clients and relative 60 virtual host running on Wowza instance.
With 20 simultaneous connections, the load average server, from htop linux shell, was about 10.00 and the streaming was very bad.
To resolve this problem we bought an amazon r3.Xlarge, but the situation Is worsened. The load average server was about 15.00 and the streaming was very very bad.
Now we have an Amzon C3.2X-large server.
With the same virtual hosts the probel is the same: the load average server is now at 21.00 and the streaming is very very very bad!
We have gone through the tuning guide.
We have not any transcoder system activated
How can we resolve this problem?
We have another question:
The values of htop linux shell, the "wowza manager" monitor and "Amazon CloudWatch Monitoriing details" are different: the first is about 21.00, the second about 40% of cpu load and the third about 25%.
What's the real value of the load average cpu server?
Following, the different Load average CPU monitor:
When you say you have 60 virtual hosts running, are you referring to 60 Wowza VHosts within a single Wowza instance?
Obviously it's not easy to perform diagnosis from here based on what we see so far. There could be a number of factors causing the high load, though in the end it could just be that you are hitting a resource bottleneck either at the CPU, IOPS or network level.
You could try running tools such as
sysstats - sar, iostat, mpstat etc. They can be left running for a while in order to capture a representative sample of what is going on at the hardware level. If you can collect this data then please feel free to
raise a support ticket (please include the information as listed on that page and any other relevant data), and we would be happy to investigate further. Please reference this thread if doing so.