Importing a binary export from SPS9 Rev 90 to SPS7 Rev 70 fails

[mert_karakilic]

I am trying to import a previously exported data from SPS9 Rev 90 into the SPS7 Rev 70 instance of HANA. It is just table import/export not a full restore. I am getting the following error;

Could not execute 'import "XYZ"."TABLE1" as binary from ...' in 2:56.686 minutes .

Data receive failed [Connection reset].

Have you seen an error like this before? Any ideas on how to fix it? At first it may seem like a connection issue but it is not. I have tried using the same import files at the same location on another instance running on the same machine with SPS9, it worked fine and imported as it should. So this issue only happens if you are importing this binary export into an older version, specifically SPS7 Rev 70.

Any suggestions would be greatly appreciated. I can not upgrade the older version of the HANA yet to SPS9. I am currently using Data Services to move the data from one instance another and that works fine but it is much slower than using the binary exports to re-import.

Thanks in advance.

[lbreddemann · Accepted solution]

Ok, wild guess here but:

SPS 9 uses L2Delta column store (internal data structure).

SPS 7 uses the old delta store implementation.

Maybe the import binary option cannot convert between old and new.

Could be worthwhile to check the indexserver trace.

In the meantime: try and export/import via CSV or SDA.

[mert_karakilic · Accepted solution]

Hey Lars,

Thanks. I am doing the CSV export as we speak, hopefully that will do the trick. I ahve not the SDA export before, I should check that.

I was able to copy the two massive tables via Data Services but one fo them is complaining about the column name length now. Apperantly we cant have column names longer than 64 charactes. Not sure why... If you data provider is crazy enough to have a 64 character long column name, you need to comply but MS SQL folks think otherwise as DS uses MS SQL in our setup.

Thanks for the help.

Mert

[mert_karakilic · Accepted solution]

Importing via CSV worked but if I had the problem with one of the other tables which is quie large both rwoswise and columns, CSV is so bad. I have not tried SDA but heard about it. Assuming the servers are in the same DC, it should be decent speed than exporting in CSV and reimporting even faster than DS, right?

Thanks for the help Lars.

[lbreddemann · Accepted solution]

That (SDA being quick) would be assumption, too. In my experience it's fairly straight forward, especially in HANA - HANA setups since no additional client software is required.