The design is in place to handle multiple jobs running at the same time and every day we have jobs running at the same time.
The problem happened because FTP from SAP BW to SDS server occurred at the very same split second . This is because all the jobs are having very less/no data and trying to connect to BW at the very same time. This Issue will be mitigated when different volumes would be processed across the jobs and its very less likely to have the failure.Also, To improve performance, some of these jobs have 4 partitions enabled in them for parallel extract so 4 ftp requests goes to BW. So if we have 2 similar types of jobs running in parallel, there would be 8 ftp request to BW and without data, there is a chance one ftp overlapping other