Sequential file Vs Hashedfile

[parimi123]

We have a server job which reads records from a Sequential file and loads the good (transformer does the validations) records into oracle table.
Apart from loading into oracle table, we also write all the good records (1015 bytes) into one Sequential file. This sequential file (good record) is used in the next job based on some conditions but its very remote.

So Instead of writing entire record (1015 bytes) into sequential file, i am planning to write a key (15 byte number) into hashed file.

In this regard i would like to know, If writing the key into hashed file instead of writing the record into sequential file impacts the performance of our application?. Does it reduce the I/O problems we are facing?

Please advice me.

Thank You,
Poorna

[alanwms]

If you only need the 15-byte key, why not write just that field to the sequential file? Your performance should always be better with the sequential file output rather than the hash file. It also seems that your sequential file output becomes the input to a downstream job. If this is the case, then you would only have the 15-byte key as the primary input. If this seq file is actually used as a secondary input, then you'd be better off creating the hash file in the upstream job and using it as the lookup in the downstream job, since a sequential file isn't a good choice for a lookup.