Injecting data from non-sap linux systems into AWS Hana - options

[former_member182500]

Hi,

I'm trying to determine what options I have for injecting data into AWS Hana from non-sap linux systems.

Its a real shame Odata PUT method is not supported in XS yet.  Grateful for indications of options available.

Many thanks.

[former_member182500 · Accepted solution]

I came across the very interesting post by Henriqe Pinto regarding use of server-side javascript in xs to perform inserts with prepared statements:

How to insert data in HANA through JavaScript (... | SCN

I do wonder how the Java Vm would scale up if being triggered with hundreds of http calls per second?  Preferable surely to packet up hundreds of records into an Odata PUT? (which is not supported as of SP05 I understand).

[thomas_jung · Accepted solution]

Although the OData service generation framework does not support put in sps5 (it is planned for sps6), you can still create PUT OData/REST services with server side JavaScript.  You can absolutely batch up records for update/insert within XSJS, so you wouldn't need hundreds of JavaScript VMs per second. Even so, we do scale VM instances well. We keep a pool of VM instances and pull needed threads from this pool.

[former_member182500 · Accepted solution]

Thanks for heads up on odata put for sps6, any rough timeline for that?

I must be missing something with regard to the batch scenario -I have seen you example regarding prepared statement batch, but the data is generated internally within the xsjs.

Assume a scenario where I have 100 records of 10 fields each, sourced from an external system to be injected into hana.  Without Odata put with its single payload, and using xsjs to insert as the alternative, an option is to call the xsjs 100 times with 10 parameters or 1 time with 100 parameters if I batch up source side.  My assumption is based on my understanding of Henrique's post and your addition. 

Perhaps you mean I can still pass a mass entity payload with the HTTP PUT but the payload is not automatically transformed or made persistent by the oData framework?

I'm also interested in how credential authentication could be done from a source system, basic authentication with user name and passowrd value pairs in the http header?

Thanks again.

[swapan_saha · Accepted solution]

Hi Jon-Paul,

While Thomas is responding to the specific question in this message, a general release related update. Whenever HANA 1.0 SPS6 (appliance) is available, you will not get SPS6 at the same time at HANA One as it goes through several additional steps to release the same software in the cloud. There could be 4-6 weeks delay between appliance and cloud version. By the way, you probably aware that HANA SPS release cycle is  every 6 months.

Regards,

Swapan

[thomas_jung · Accepted solution]

SPS6 is planned for general release in the June time frame. As stated, it does take several weeks following the general release for HANA One to be upgraded.

>Perhaps you mean I can still pass a mass entity payload with the HTTP PUT but the payload is not automatically transformed or made persistent by the oData framework?

Yes absolutely.  You can fill the body of the request object for anything more complex than what can be passed easily via URL parameters. Generally people will use a JSON body for the request because that is so easy to work with in Server Side JavaScript. 

>I'm also interested in how credential authentication could be done from a source system, basic authentication with user name and passowrd value pairs in the http header?

Your authentication choices are SAP Logon Ticket (acceptance only), Forms Based Authentication, Basic Authentication, or Anonymous services (service user configured via SQLCC). In SPS6 we plan for SAML and X509; with OAuth and Kerberos coming further down than line than SPS6.