Last time I participated in #SAPChat on HANA and shared by experience here . Fortunately, I could participate in the chat on HANA as held on this past 12th July at 9:30 PM IST. Timing is very favorable and I could not afford missing it. The purpose of this blog is to share only the questions as asked by me and the corresponding answers. In the last chat, I added @SocialKev to the list of people I follow, and this time I added another two awesome people as @tweetsinha and @D_Sieber . Answers provided to my questions were not good rather great. It is good to follow Mentor @tpowlas as she participates in every chat and asks questions which could possibly be in everyone’s mind which further encourages us to participate in the chat.
Without further delay below are my questions and corresponding answer:
Q1: Why so much noise on Big Data? We already had many efficient data mining algorithms?
- Real time on big data is useful, big data in itself is for few e.g. Real-time re-pricing. Time traversal is uniquely possible on SAP HANA. Big data is just one of the five dimensions from @vsikka. [I would have to find out the other four. ] In @applebyj’s opinion HANA is not big data. Big-ISS data. TB not PB. Real time is a good use case. [No wonder a lot depends on the use cases for HANA and I would have to read blogs and other materials available for them here on SCN.]
Q2: With HANA, not only hardware but also ABAP changes, then why is it said to be non-disruptive?
- Non-disruptive as e.g. COPA end users don’t see disruption but only faster results. HANA platform has no versions. [I replied that HANA platform might not have versions but HANA studio certainly has many versions and that too very fast but then I got many tweets justifying the same]
Q3: Any specific algorithms planned to analyze big data on HANA?
- Yes and I am advised to check Predictive Analytics Library for SAP HANA. @narenchawla says popular clustering, classification, regression etc. are implemented as part of Predictive Library in HANA. [I would have to check the library to understand all the terms though.]
Q4: HANA does not solve all the problems in real time then, why is it said to be real-time processing?
@applebyj says because with HANA data can be provisioned in real-time and reported without aggregation and layers. @tweetsinha says we say real-time as new class of applications where OLTP + OLAP is in one database, lot of start-ups and new applications using this. @vijayasankarv further questioned me if it’s a use-case or HANA problem to which I replied that I had thought real-time to be a misnomer but with the answers I am clear.
Q5: Errors thrown in HANA studio does not give any clue to the problem and is very unstable?
@NathanOyler says error messaging is getting improved with every version of HANA studio. @vijaysasankarv says studio is getting revamped and inputs are being taken from many people in ecosystem. @tweetsinha suggested reporting all such issues to askSAPHANA@sap.com. I have planned to report the below: [anyone having issues to be added, feel free to comment in the blog]
a) At times while uploading data through FTP data is only uploaded till particular no. of records. After sometime the error gets resolved on its own. Not sure if the same happens with newer techniques of data upload.
b) Coding while creating procedures and calling views inside it is a big pain as I find it to be very inconsistent. Error thrown gives no clue to the problem.
c) Primary keys defined in table do not automatically get reflected while using them in attribute views. This can cause inconsistency.
d) HANA studio at times acts very unstable and does not show the error after sometime for the same code.
e) While creating calculation views, the same parameters appear on the right hand side as ‘Calculated Columns’. However, both play very different role. Some documentation should be available highlighting the difference and similarly for other parameters too.
Q6: Any Online event planned like Online Innojam we had months before for ABAP on HANA?
I am very satisfied with the answers and a big thanks to the people who answered the queries. I really wonder how do they track all the questions asked and answer each one within a time-period which often fall short for HANA discussions.