System replication async mode

[Former Member]

Hi All,

I have a very strange doubt here.

We have two data centers A and B. When we do system replication between A and B the speed we observed is about 60 MB/s.

When we scp a large file from data center A and B we get about 40 MB/s.

I wanted to understand the mechanism as to how HANA does the system replication.

When we do the system replication from A to B in async mode then data from A disk goes into B disk right and then it would be read into in memory of B right.

How is that operation of system replication is faster than transferring a file.

We have done scp file into B using /dev/null and also new file creation but we see that the speeds are slower than what we see in SAP hana system replication. So wanted to understand as to how the system replication happens.

Could you please help me understand this.

[Former Member]

We were able to test, it was due to large master indexserver data file size.

[Former Member]

Hi,

It is not clear, could you elaborate?

Do you mean master indexserver data file size is responsible for the difference in speed of a scp operation in comparison with the system replication data transfer?

Sunil

[Former Member]

Hi Sunil,

Ultimately the snapshots are transferred. If the master node is extremely large then inspite of using dedicated replication networks it wont help.