I believe there may be something in the new release (SPS07) that is causing other issues. I now have a ODBC issue with Data Services and will be opening a OSS message on SAP. I would prefer not to revert to the previous version and currently troubleshooting this issue on a Windows Server with SQL Server 2008. The HANA system is on SuSE 2.6 and communicated without an issue with release SPS06.
I understand from a colleague of mine that the high asynchronous read/write ratio may be an indexing problem. With that in mind, do you have any aggregate (wildcard *) searches in place? I remember the MARA table being very large on ABAP - was it indexed before you ran the SLT conversion to HANA?