@MAHESH_P

I'm assuming you are referring to custom properties here.

Couple of questions -

1. How do you know that no of columns of work table is causing the performance issue, and not the no of records on that table?
2. Have you checked with DBA or generated the explain plan of query with performance issue to verify the root cause of the performance issue?
3. Have you checked the average size (in bytes) of rows of that table? Is that too high?

Options -

1. PRPC always keeps the data in pzPVStream/BLOB and copies the same to the column of an exposed property. You can drop those columns from table, reset DB connection with PRPC nodes (by saving Data-Admin-DB-Name instance). If exposed columns are required for reporting or search, this solution won't work.
2. Modify your data model - It sounds like you are keeping too many properties directly at your work class level (unless you're using external mapping on class def to bubble up properties at work level), which is not good from data model design perspective. Ideally, you should have a better data model of work object where such custom properties should be moved to appropriate sub level under pyWorkPage (using embedded pages/list). Then use declare index to expose them on declare index table (separate from work table). 

Thanks