Qlikview Cluster & Reload

[#1]

Quick Question

If we have cluster environment.

• 2 QVS machine in cluster
• 2 Publisher machine in cluster

Who handles the reload. is it the server that handles the reload or the publisher.

My assumption is that Publisher service is Qliview distribution services, and it starts QVB.exe for each reload process that initiate. (is this true). This has nothing to do with QVS cluster and its cpu or memory. The reload will use publisher cluster cpu and memory.

[Anonymous]

MD

Yup, you are correct.

The reload will use publisher cluster cpu and memory.

Best Regards,     Bill

[its_anandrjs]

Hi,

Yes right basically Server manages the Reload task, application and access and Publisher publish and distribute the Qlikview applications and do reloads

Hope this helps

Thanks & Regards

[#4]

Are you sure Aanand. I don't think so.

Server manages application and user access.

Any reload from sources is handled by Publisher.

Can someone please verify this.

[#5]

anyone with clarification.

I was under the impression that all reload jobs where qlikview uses connections, tables, SAP etc is handled by publisher. and if publisher is on separate cluster from QlikView Server. publisher reloads the data and the publish the output final qvw to QVS user folder.

[Miguel_Angel_Baeyens]

Hi,

We are mixing up commercial names and services, which is not always correct. The QlikView Distribution Service is the service that perform all tasks, even when there is no Publisher license.

The QDS is able to reload, reduce and distribute to either a file system folder or a QlikView Server.

Miguel

[#7]

Hi Miguel

Clarification:

Lets say we have 2 QVS.exe cluster and 2 publisher cluster

QVS.exe cluster will hold the user application which was published by publisher cluster to USER FOLDER and loads it into memory upon request by users. Basically it serves the access point with all the published applications.

Publisher cluster is the one which runs reload task, fetches data, creates final qvw. The publisher cluster uses distribution task to publish that app to QVS cluster.

is above assumption correct?

[Miguel_Angel_Baeyens]

Hi,

Yes, that is correct. To be a bit more precise, what QlikView Server service does is handling all the data models in memory, do the aggregations in the memory and cache them, unload documents where timeouts happen, and so on.

Miguel

[#9]

how do you think virtual servers handles reloads task. do you see virtual being any different then physical as reloading cluster.

what is the way to determine your publisher server CPU & Memory. The way I am figuring out is keeping an eye on QVB.exe that runs and the resouces it takes which dictates work publisher is doing. Am I correct in my thinking.

[Miguel_Angel_Baeyens · Accepted solution]

Hi,

There may be indeed differences between virtual and physical environments. For example, when the virtual environment is sharing its resources with other virtual machines.

The way to size the server for the distribution service, is that it needs in memory the sum of all the apps that are going to be reloaded at the same time. Let's say you have to reload every day 4 QVW files that use about 1 GB RAM each. Then you will need at least 4 GB RAM only for QlikView.

CPU will vary depending on the time frame you have for your reloads, meaning that if you need to reload all your apps in 3 hours you will need to add more resources than if you have 12 hours.

Anyway, I'd recommend you to peruse this documents QlikView Architecture and System Resource Usage and QlikView Server Memory Management and CPU Utilization in Technical Briefs | Qlik that will help you to better understand QlikView sizing and engage with your local Qlik Consulting Services team.

Miguel