1. Purpose
   1. Load Empty Table with no secondary indexes
   2. Execute on Channel,network attached and NCR server
   3. Checkpoints for restart
   4. 2 Phase Loading
      1. Acquisition Phase (Phase1)
      2. Application and Insert Phase (Phase2)
   5. Error table on both the phases
   6. Status of run is displayed
2. Restrictions
   1. Can load only one empty table by 1 FL job
   2. Can accomodate 15 FL/FE/ML script at one time
   3. Cannot load table with Referential integrity,secondary indexes, JI, Hash indexes and triggers
      1. Can load table with soft referential integrity
   4. Duplicate rows cannot be loaded on multiset table (Use multiload to load duplicates on multiset)
   5. If amp goes down, FL cannot restarted until it is back online
   6. Don't use fastload input file from NFS mount
3. 2 PHASE Loading
   1. Acquisition Phase
      1. Step 1: FL is executed, sends blocks of records to PE session
      2. Step 2: PE receives records block and send to AMP via BYNET
      3. Step 3: AMP receives block of records in memory
      4. Step 4: AMP rehashes each record and redistribute through Message passing Layer (PDE and BYNET)
      5. Step 5: Message passing layer passes records to appropriate AMP based on row hash value and store it in memory
      6. Step 6: when enough rows collected to fill block, then written to the disk
   2. Insert or Application phase
      1. Step 1: Fast load receives 'END Loading' statement
      2. Step 2: Send request to PE for phase 2
      3. Step 3: PE broadcast about start of Phase 2 to all amp
      4. Step 4: AMP reads its block from disk
      5. Step 5: Data record is sorted based on row has sequence
      6. Step 6: Write sorted records back to disk
      7. Step 7:If fallback is defined, fallback rows are added
4. Supported Input format
   1. Binary
   2. Formatted
   3. Unformatted
   4. Text
   5. Variable-length text