Sybase IQ and HANA. Insert as select speed.

[Former Member]

Dear community.

My results for mass inserts.

Environment

Sybase IQ (15.4 ESD2)       - 8CPU Xeon (2009), Hi-End array.

HANA Box (SPS5, Rev 45)  - 40CPU E7 (2011), midrange array.

TABLE TST - (87 columns), 145 mln rows. Primary key - 6 columns.

Results:

IQ - 43 mins

HANA - 50 mins.

[lbreddemann]

Mikhail,

sorry, but given the information in your posts you could also just write:

IQ fast - HANA slow.

Would be the same level of information.

We don't know about your test environment and the purpose of the tests.

What are we supposed to do with this kind of information?

That kind of comparison was worthless with Oracle vs. MS SQL Server vs. DB2 vs. mySQL ...

Why should it be useful now?

just my 2ct on this approach.

Lars

[Former Member]

Lars.

Yes, it's my personal experience, and i am share with others.

Trust or not - everybody must decide at self.

[Former Member]

Mikhail, it is great that you share your personal experiences with others, and I fully agree that everyone should decide for themselves, based on the data available.

That said, you do not have enough data presented here for others to decide based upon.

I routinely see customers loading hundreds of millions of rows of data in much less time than you are presenting in your results here. In fact, I am right now looking at a summary of testing results where a customer loaded 530 million rows of data into a single-node HANA appliance in 28 minutes. And I am familiar first-hand with results many times faster than this.

So, as Lars suggested, if you are going to post such results, it would be helpful to others who must decide, to understand more about your environment, both hardware and data model, and the type of testing you are performing. It would also be great to know the requirements behind the testing, for contextual understanding.

Cheers,

David.

[Former Member]

David,

HANA Box - 512GB RAM, 40CPU.

Just one table TST  - in it  FI-items from ERP (0FIGL_4 extractor). Tables about 145 mln rows, 87 columns.

insert into TST2

select * from TST

Please ask more question.

And one more thing - everyone who have HANA can compare by your self. Dowload IQ trial - it's free.

[Former Member]

did you try to compare the performance of the select statement only as well?

[Former Member]

Thank you, but this is all information you have already provided.

[alexander_kutz]

I like the test - it gives a impression. I mean comparing HANA with an old, established and well-engineered software gives a first impression of the development status of HANA. SAP is pushing HANA, but you can say what you want - it is a young product. And it takes time to perfect it.

[Former Member]

Now, trying the mass inserts in row_store type of the table

1. Creating TST2_ROW as copy from TST2. But select type  - row store table,

2. Execute

    insert into "BWUSER"."TST2_ROW"

    select *

     from "BWUSER"."TST2"

Comments:

CPU usage less than 1%.

[Former Member]

TST2 - still 173GB on disks...