Skip to Content

Materialization in HANA

In one of my client they are proposing to materialize everything e.g all views and upload result set into table. However they go beyond that even they materializing all source tables with almost no logic into HANA tables. We generally materialize on demand basis with extreme scenario where to resolve join issue on calculation column. Also they are loading data based on SLT load date time e.g. delta load based on when last time the SLT updated. They have scheduled jobs every 15 minutes to run 100 of jobs to load delta records. What could implication on HANA system if this jobs run in background when people query the system. How HANA differentiate between these two processes. They are currently facing issue with multiple sources (some views having 17 tables, which time stamp they can consider). Now we are thinking about listing down all anticipated issues and I need help on the same. Can you please what could be the issues for this kind approach. They seems to pursue the data vault technique in conventional data warehouse as it is SQL based and easy to read, understand and troubleshoot for people with understanding of SQL with less HANA Modeling knowledge. Also I would like to understand the impact of this approach in HANA 2.0 SPS3. What they can do and can not use features available in that version.

Add comment
10|10000 characters needed characters exceeded

  • Hi Dipanshu,

    It is unclear what your question is, the scenario and the problem. There are at least three topics here starting by materialization of results, replication strategies and I cannot pin down the third with the description you give

    I suggest you explain the three different problems clearly in different questions with what the intended result is, what you have tried and researched and what is failing. Please also take the time to make your question clear and easy to read if you expect someone to spend time on it and help.

    Regards,

    Lucia.

  • Get RSS Feed

4 Answers

  • Best Answer
    Nov 07 at 12:02 AM

    Nobody here knows your project, your data or your requirements an yet, you ask for potential issues of how you implement your system.
    You even throw in a data modelling methodology (data vault) that is independent from the DBMS used to implement it.

    I guess you can see how this line of questioning is at odds with what can reasonably be done in a Q&A forum.
    What you seek is the advice of an experienced data warehouse developer for your project - not some "how to" tips from some people on the internet.

    Since you asked for opinions on some topics, here are mine:

    As general direction, when dealing with Data Vault it's probably best to go back to the source http://danlinstedt.com to learn all about it.

    As for materialisation of result sets and regular refresh: it's highly likely that this approach is not getting the best value out of the HANA investment. Most HANA based scenarios explicitly avoid materialising result sets so that performance and flexibility can be retained and operational efforts for keeping the pre-computed results current can be minimized.

    Good luck with your project.

    Add comment
    10|10000 characters needed characters exceeded

  • Nov 06 at 03:59 AM
    -1

    Hi Lucia,

    Let me clarify my first query which is if we pursue materialization of result for each model we built in HANA, then what could be the issues in future. The materialization happens though running procedure in every 15 minutes by uploading views into table in HANA schema irrespective of any performance issue. My questions would be following,

    1) If there are around 100 procedures running every 15 minutes for HANA models , how it could be affecting HANA system performance.

    2) If we materialize everything in HANA, what we would be missing apart from real time reporting

    3) The delta load of procedures is based on SLT update/insert timestamp.What could be the potential issues if we ignore issues related to SLT refresh.

    4) Also client seems to be pursuing data vault methodology in HANA, is HANA db suitable for data vault mehtodology?

    5) What could potential issue for data vault methodology in HANA 2.0 SPS3.

    Thanks & Regards,

    Dipanshu

    Add comment
    10|10000 characters needed characters exceeded

  • Nov 08 at 06:58 PM

    Hi Lars,

    Thanks for taking time to read my query and answering the same. I was not looking for specific details on data vault but just tried to provide context of materialization in this project as client is trying to follow data vault methodology in SAP HANA. Thanks for providing your insight about materialization in HANA . As it is new to me as well , I am trying to understand the system impact of this kind materialization and finding out in SAP world , if anybody heard or worked on data vault (materializing all result sets) and shared their experience for the same.

    Add comment
    10|10000 characters needed characters exceeded

    • "data vault (materializing all result sets)" - that's not data vault at all.
      Sure you can call the approach of materializing all result sets whatever you like - but this has zero connection to the DV modelling approach and method.

  • Nov 09 at 05:00 PM

    Hi Lars,

    Yes you are right , as per the methodology the raw vaults, business vaults ,satellite, link , link satellite , hubs are being materialized. But I was not trying to emphasize on the same as my main objective was to find out the issues on materialization which people might be facing in HANA.

    I hope that clarifies. Thanks for your response.

    Regards,

    Dipanshu

    Add comment
    10|10000 characters needed characters exceeded