compression of a cube

[Former Member]

Hi Gurus,

I am supposed to carry out compression on a cube. Everyday the loading to the target occurs and later indexing happens. The compression has to be done for requests since an year from today i.e. october 2013 till today. Should I wait until the loading and indexing happens or can I directly do the compression right away, and also I am supposed to do compression for 20 requests at a time and each request contains around 300,000 records. How can I do that?

Regards,

Rajesh.

[Former Member]

Hi Rajesh.

I ran the test in the sanbox system and found that you can load and compress at the same time.

I believe, You can run few tests at your end too and proceed with the activity safely.

Regards,
Saurabh Ojha

[former_member197660]

Hello Rajesh

As mentioned by Raman you need to fist compress the cube with all historical request manually to avoid initial long runtime of the chain.

Once the manual compression activity is completed you can add the compression of cube step in process chain as the last step, do not try to compress the data in parallel with data loading or index creation.

In general the decision to keep the uncompressed requests depends on the application and the necessity of data for upper layers, but in case you are not sure then its advisable to keep uncompressed data for 1-2 weeks.

Also you can create a maintenance process chain which would compress the data of infocube and schedule it on weekends to avoid any issues.

Regards

Ashish

[former_member186399]

Dear Rajesh,

1) In a process chain compression is done before index creation. As you are doing for the first time the compression you can do it if there are no loading taking place in the infocube.

2) Please note that the cube gets locked when you do the compression and it can impact you reporting. What you can do is check for the time taken for compression of 1 to 5 request and then according plan your activity. You can compress keeping at least 30 request .

3) Once your done with the compression of all the request and then you get it added to you process chain before the index step. You cannot do simultaneously indexing and compression or loading or compression

Regards

Gajesh

[former_member182997]

hi

Rajesh,

I don think so as while you will compress the cube you will access the f table which will also be accessed by the scheduled process chain while loading the data. This may generate  a Lock issue.

this i think is wiseful

deletion of index-->loading of data--> creation of index-->compress the data
As data volume is large, please take care to carry out in non business hours

Lets see what experts say

[RamanKorrapati]

Hi,

Please don't compress till today requests.

Better to compress the request which are older than 2weeks. mean upto Sep 10 2014.

After finishing of present loads, you can start compressing 4 or 5 request at once.

Thanks

[Former Member]

Hi Ramanjaneyulu,

You mean to say to compress requests from oct 2013 till 10th sept 2014 and then once it is done and todays loading is done we can compress 5 requests at a time. My question is while compressing is goin on can the loading be done simultaneously today and also can we do compression while the indexing for the target is in process.

Regards,

Rajesh.

[RamanKorrapati]

Hi,

My question is while compressing is going on can the loading be done simultaneously today and also can we do compression while the indexing for the target is in process?

No and not allowed.

If there is no loading to indexing on cube then only you can do compress.

if those are using thru loads or some other actions compressing may fail.

Thanks