How to calculate metric Core Qlik Replicate

[calonsca]

Hello,

I was looking at the documentation on the "Core" metric ( https://www.qlik.com/us/-/media/files/legal/license-agreements/additional-terms/qdi-gold-client-and-... ) and now I have the following doubts regarding how many Core I would need.

• Is the amount of Core related to the amount of tasks I have in  Qlik Replicate ? I mean, is having 20 tasks connecting to 20 Oracle the same as having 200 tasks connecting to 200 Oracle? As far as number of Cores is concerned
• If I have a server with 8 cores, how does it relate to the Core metric in Qlik Replicate?

The questions may seem a bit foolish, but I don't quite know how to calculate the Core metric I need.

Regards

[lyka]

Hello,

As a start you can refer to this link - there is no computation but you can use it as reference:

https://help.qlik.com/en-US/replicate/May2021/Content/Replicate/Main/Installation/Recommended_Hardwa...

Thanks

Lyka

[calonsca]

Hello Lyka,

In that link what I can see is the hardware level configuration for the Qlik Replicate server, how does it relate to the "core" metric for extractions ?

[Bill_Steinagle]

Hello,

Thank you for the post to the QDI forums on the Replicate environment and some of the metrics about the environment. Some information below to consider when you are setting up a new Replicate environment and the Memory and CPU utilization. 

- each Task we have the below internal settings for the thread and size 3x8 = 24MB (note these are not in the JSON of the Task) default limits

"stream_buffers_number" : 3
· "stream_buffer_size" : 8 (MB)

- each Task can have up to 5 connections on the Source with the one being active other idle

- 20 Tasks 24MB in size and 100 threads (20x5) this can help with the sizing of the environment

- 200 Tasks we donot recommend but 200x5 along with the 200x24MB for the sizing of the Replicate environment

- 200 Tasks per Replicate Server not recommended due to the Heap Sizing on Windows OS limitation even though 64 bit Product

Hope this information will help you find the information on the Replicate environment and the provisioning of the Replicate Server.

Thanks!

Bill

- if you setup more than 100 Task it is recommended to create a new Replicate Server 

[lyka]

Hello,

with regards to : 20 tasks connecting to 20 Oracle the same as having 200 tasks connecting to 200 Oracle"  , 'Replicate establishes more than one connection when connecting to a n Oracle source so you should also take that into consideration.

The number of connections vary by endpoint and may in fact vary by specific release as new requirements are implemented or optimizations are applied.

As an example please see a list of potential connection one might see for an Oracle endpoint.

1 X metadata connection for a task

Full load:

2 connections per a used subtask (unload and LOB)

CDC:

1 LOB connection

1 SQL request connection

1 Deferred Constructor connection

If BFILE is in use – one connection per RAC node

If ASM is in use – one ASM connection per RAC node

If using parallelASMReadThreads, for ASM, you would multiply it by the number of parallelASMReadThreads.

Hope this helps!

Thanks

Lyka