TDP for SAP HANA and "Server out of data storage"

Rigido

ADSM.ORG Senior Member
Joined
Apr 21, 2006
Messages
151
Reaction score
7
Points
0
Location
Rome, Italy
PREDATAR Control23

Hi all,
maybe it is working as designed but I think there is a design issue:
The problem is that when you try to backup a SAP HANA instance on a 6TB RAM appliance TDP will first ask to TSM/SP server for the whole 6TB space despite the "real" space occupied by the instance.
We moved from DISK devclass to Directory (just to let you know, great compression and deduplication ratio on HANA LOG pools, no deduplication on HANA DB pools) but we still need 6TB of free space on filesystems.

Did someone face this issue?!

Thanks
 
PREDATAR Control23

Hi Rigido,
did you separate the utl files for HANA db backup and HANA log backup already?

f.g.
global.ini content :

[backint]
data_backup_parameter_file = /usr/sap/SID/SYS/global/hdb/opt/hdbconfig/initSID.utl.db
log_backup_parameter_file = /usr/sap/SID/SYS/global/hdb/opt/hdbconfig/initSID.utl.log

[backup]
data_backup_parameter_file = /usr/sap/SID/SYS/global/hdb/opt/hdbconfig/initSID.utl.db
log_backup_parameter_file = /usr/sap/SID/SYS/global/hdb/opt/hdbconfig/initSID.utl.log
 
PREDATAR Control23

Hi Rigido,
did you separate the utl files for HANA db backup and HANA log backup already?
Hi Cilesiz,
thank you for your answer but why I should do that?
I have just one .utl file and different management classes for DB and LOG.
Each mgmt class with it's own archive copy group and different storage pools.
 
PREDATAR Control23

I notice that you said you were getting no deduplication on the Hana DB. We also used to get none and logged a call with IBM. Turns out there was an issue which is fixed in 7.1.7.
We do now get deduplication of Hana DB.

This is one of our Hana nodes - stats for logs and DB combined.

Code:
Protect: HRNTSM01>q dedupstats hrn_cont HRNFHTB01-HANA

Date/Time              Storage Po    Node Name     Filespace     FSID    Type    Total Saving    Total Data Pro
                       ol Name                     Name                            Percentage       tected (MB)
-------------------    ----------    ----------    ----------    ----    ----    ------------    --------------
02/21/17   09:22:32    HRN_CONT      HRNFHTB01-    /tdpmux          2    Arch           87.69         2,906,458
                                      HANA

Protect: HRNTSM01>q stg hrn_cont f=d

                     Storage Pool Name: HRN_CONT
                     Storage Pool Type: Primary
                     Device Class Name:
                          Storage Type: DIRECTORY
.
.
.
            Total Unused Pending Space: 229,028 M
                 Deduplication Savings: 19,759 G (47.11%)
                   Compression Savings: 9,284 G (41.86%)
                     Total Space Saved: 29,043 G (69.25%)

There is a mix of different data in that pool but most of it is Hana database backups.
 
PREDATAR Control23

If you' re using incremental or differential backup and you are using a single Utl file the Backups will go to the Log stgpool.
Right now we are going for a full daily backup but thank you for the info.
Just to let you know, "sizing" of backup object is a little bit tricky with TDP and SAP HANA.
As a quick response backint will create one file per service (nameserver, indexserver, etc.) and it will ask TSM for "appliance's RAM"/2 for every file.
 
PREDATAR Control23

I notice that you said you were getting no deduplication on the Hana DB. We also used to get none and logged a call with IBM. Turns out there was an issue which is fixed in 7.1.7.
Hi DazRaz,
which component should be at 7.1.7?
I.E. our HUP instance uses TIVsm-API64-7.1.4-4 and TIV-TSMERP-HANA-7.1.3.0-0 on a 7.1.7.100 backup server, and these are fresh generated dedupstats for DB data pool:
Code:
tsm: BCKSERVER>q dedupstats HANA_PROD hup f=d

  Date/Time: 05/05/2017 14:32:13
  Storage Pool Name: HANA_PROD
  Node Name: HUP
  Filespace Name: /tdpmux
  FSID: 5
  Type: Arch
  Total Data Protected (MB): 15,353,197
  Total Space Used (MB): 10,072,483
  Total Space Saved (MB): 5,280,714
  Total Saving Percentage: 34,39
  Deduplication Savings: 0
  Deduplication Percentage: 0,00
  Non-Deduplicated Extent Count: 64
Non-Deduplicated Extent Space Used: 24,432
  Unique Extent Count: 6,041,490
  Unique Extent Space Used: 16,098,993,351,858
  Shared Extent Count: 2
  Shared Extent Data Protected: 130,870
  Shared Extent Space Used: 65,435
  Compression Savings: 5,537,229,890,936
  Compression Percentage: 34,39
  Compressed Extent Count: 4,657,010
  Uncompressed Extent count: 1,384,546
Did you enable RL_COMPRESSION and MULTIPLEXING on .utl file?

Thanks.
 
PREDATAR Control23

Finally we were able to apply the workaround that IBM told us: https://www-304.ibm.com/support/entdocview.wss?uid=swg21694831

Setting the variable ESD_HANA_PREDICTION_FACTOR to 6 in .customer.sh file on the home directory of the SIDadm user, backint tried to send a 1009GB file instead of the 2959GB.

As this is a per database setting it is ok for us to deal with it...waiting for a bigger VTL.
 
Top