Thanks for your response Johannes,
We are having only two servers as you said one server will have both QVS server and the Publisher and the other one will have only the Qlikview server.
I have few Queries like
Is it possible to happen ---------> Qlikview Server service is down but still Publisher work properly?
If the publisher is down still users will be access the QVW files(Which are already reloaded) through web?
Ok, so machine A runs QVS + Publisher while machine B runs just QVS. That works fine.
If the QVS Service goes down on machine A, Publisher should still work, depending of course on the reason that the QVS is down. If the machine crashed then obviously Publisher is not working etc.
Also, the QVS nodes are clustered so if one is down the users will get redirected to the other one.
If Publisher is down, of course the users can still access the files that have been distributed to QVS, but no new refreshed material until Publisher can perform said refresh.
The accesspoint performs load balancing between the QVS nodes and will sense when the QVS nodes are not responding so you'll get directed to one that is up and running.
There's two types of load balancing that can be set from the AccessPoint settings in QEMC, Random or Loaded Document. Random is a round robin type load balancing where every new request goes to the next node. Loaded Document is a bit smarter and checks if a QVS node has the requested document loaded in RAM already and tries to direct a user there. If one only uses one document though this means that all users would end up on one node which is not preferable.
Once again thank you very much Johannes.
It is very useful but there are few things which i got confused. Since the Clustering is mainly to maintain the High Avilability.
1.In QlikView if Machine A and Machine B are in clusters then both the nodes are active or One is active and the other will be passive. And when the Master is down the Slave will be active i.e Machine A is passive and Machine B will be active.
2. Consider the QEMC has setting as Random and current user request has to be sent to Machine A and the machine A is down what will happen to the request or the access point will check the node is up or not then transfer the request to the respective node.
3.How the configuration details are shared between the clustering server.
4. Will there any specific changes has to be done when the single sign on is implemented.
I am trying to find these details from the Server reference manual.but still it will be helpful if you share the same.
1. Both nodes are active. However, the AccessPoint can sense if one of the nodes is not active, and direct traffic to the other node instead.
2. See above. AccessPoint will notice that Node A is not responding and direct traffic to Node B.
3. When you cluster two or more nodes these will be configured as one unit in the QEMC.
4. Single sign-on is a whole chapter in itself depending on how it's being implemented. NTFS with AD? Third party authentication? etc.
My limited understanding of load balancing using AccessPoint is that, from a high availability standpoint, the AccessPoint itself becomes a single point of failure.
If you have only 2 physical servers, and you run AccessPoint on one of those two servers, if the AccessPoint becomes unavailable, the whole system is unavailable, until an operator intervenes to start up an AccessPoint on another server.
According to this QlikView white paper on clustering QVS, http://tinyurl.com/3jtg6ps, you can avoid the AccessPoint single point of failure by implementing load balancing with highly available SW or HW based network load balancing, providing certain capabilities like session affinity are supported.
Please correct me if I am wrong.
Also, this post is somewhat related to my other recent post, http://community.qlik.com/thread/30135, regarding Publisher load balancing and high availability. I'd greatly appreciate some insight there, too.
You can implement 2 or more accesspoints if needed. You will only need a router thats capable of
1. Supports ‘sticky sessions’ – this ensure a users session persists on the same node
within the cluster, usually by using a cookie to achieve this.
2. Availability – the load balancer checks the availability of the AccessPoint web server
and QlikView servers.
3. Some form of load balancing algorithm to determine which server is least loaded.
Most companies, at least companies big enough that they might be interested in high availability, would probably already have this kind of routers.