Connect load balancing

4 replies
1328 views
Translate

Hi all, I was wondering if anyone would be able to explain, or point me to a resource how Connect handles the load balancing when deployed in a cluster? I can't seem to find any documentation on it.

Thanks

This topic is closed to new replies. Start a new post to keep the conversation going.

Correct answer

Billmam

You may want to look at the content of this

http://www.connectusers.com/tutorials/2009/04/clustering_nlb/index.php

I read it a few months ago, i believe it gives a high level view of loadbalancing. I am in the process of configuring on our network with 2 in a cluster, but we are using LBR with Apache+Stunnel and 2 Connect Servers.

the license you have must be a cluster license, otherwise i dont think it will work. the reason we are using Apache and Stunnel was due to using SSO.

but high level to answer you question, it is the database that keeps track of the clustering and does a roundrobin with the client users between the Connect servers. the 2 connect servers point to the same sqlserver database, but the content data is replicated on all Connect servers. Connect does not yet support the use of Network storage between a clustered server.

also your cluster must be on the same subnet due to the replicated content data.

hope this helps some, if you have more questions i will monitor this thread.

A

Anonymous

Correct answer

Billmam

You may want to look at the content of this

http://www.connectusers.com/tutorials/2009/04/clustering_nlb/index.php

I read it a few months ago, i believe it gives a high level view of loadbalancing. I am in the process of configuring on our network with 2 in a cluster, but we are using LBR with Apache+Stunnel and 2 Connect Servers.

the license you have must be a cluster license, otherwise i dont think it will work. the reason we are using Apache and Stunnel was due to using SSO.

but high level to answer you question, it is the database that keeps track of the clustering and does a roundrobin with the client users between the Connect servers. the 2 connect servers point to the same sqlserver database, but the content data is replicated on all Connect servers. Connect does not yet support the use of Network storage between a clustered server.

also your cluster must be on the same subnet due to the replicated content data.

hope this helps some, if you have more questions i will monitor this thread.

Translate

A

Anonymous

Thanks for the reply Jimmy, that did help alot. I was just wondering though, is there anywhere that describes how Connect defines the failover in the round robin? Or if theres anyway to customize it?

Thanks again

Translate

A

Anonymous

I have not found any documents, only information i have gotten from the Adobe Tech support group. The Connect documents are too high level and make too many assumptions. and no real way to customize with 2 servers. you could only set the order by using the custom.ini for the RTMP. if you look at the config.ini file you can look at options that can be changed.

another idea would be to use Apache (web side) and sTunnel (rtmp) to front-end Connect. with apache and stunnel you might be able to control the loadbalancing better.

Translate

Sign up

To post, reply, or follow discussions, please sign in with your Adobe ID.

Sign in to Adobe Community

To post, reply, or follow discussions, please sign in with your Adobe ID.