Deciding which Gateway in Gatewaycluter accesses which Datasource

[T4S3R]

Hi everyone,

we have a setup consisting out of multiple Gateway instances. Each instance / server can access a different datasource but none of the instances can access all datasources needed.

As an example, one instance can access a HANA Database while another instance can access a SQL Server. Neither of them can access both datasources.

Now the questions:

- What happens if I link all the Gateways in a Gatewaycluster and try to refresh a Dataset with different datasources?

- Is the cluster smart enough to know which Gateway can access which source or will it run into an error message?

- If it runs into an error Message, is there some way to decide which gateway in the cluster should access which datasource?

Best regards and thank you in advance!

Timo

[lbendlin · Accepted solution]

- What happens if I link all the Gateways in a Gatewaycluster and try to refresh a Dataset with different datasources?

Here's what actually happens.  The anchor member (the dispatcher, if you will) looks at each data source  (NOT dataset) refresh request and asks a random cluster member to handle the request (can also ask itself).  If that cluster member does not have access to the data source then the anchor will ask another member etc.

- Is the cluster smart enough to know which Gateway can access which source or will it run into an error message?

No, it is not smart enough. It assumes that all cluster members have access to the data source (and have the necessary drivers installed).  What you are doing is make life unneccesarily hard for the anchor. You're wasting cycles.

- If it runs into an error Message, is there some way to decide which gateway in the cluster should access which datasource?

If you really need to segregate date source access, use multiple gateway clusters.  Don't torture your single gateway.

[T4S3R]

Hi @lbendlin 

thank you very much for the reply it helps a lot!

Is there a way to circumvent the problem that you can only assign one Gatewaycluster to a Dataset? Otherwise using multiple Gatewayclusters is not an option.

Best regards

Timo

[lbendlin]

"the problem that you can only assign one Gatewaycluster to a Dataset"

Why is that a problem for you ?

[T4S3R]

Hi @lbendlin,

every Gateway has individial port restrictions, meaning that the underlying Gateways can (at least right now) only access 1 or 2 Datasources at once.

The setup right now is that there are two main Datasources (lets say SQL1 and SQL 2) that can be accessed by all Gateways. On top of that, multiple other datasources need to be added (lets say SQL3 and SQL4) where only certain gateways have access.

We would need to create a cluster and therefore servers for all different combinations so one cluster for SQL1 and 2, one for SQL 1 and 3, on for 1,3 and 4 etc. 

I would like to tell you why it is done like that but i do not know myself.... Thats why i posted the initial question :).

Best regards and thanks again for your help!

Timo

[lbendlin]

I see two options

- persuade your network IT team that these seemingly arbitrary port limitations are not necessary

- use your current setup and accept the fact that it is a very inefficient use of your gateway resources.