Why do these insert vary so much in performance?

Question

I have a table and a package similar to those shown in below DDL.
Table TABLE1 is populated in chunks of 10000 records from a remote database, thru TABLE1_PKG
by receiving arrays of data for three of its fields and a scalar value for a set identifier
in column named NUMBER_COLUMN_3.
I have two databases with following record count in the table:
     DATABASE1: 55862629
     DATABASE2: 64225247
When I executed the procedure to move 50000 records to each of the two databases, it took 20 seconds to
populate DATABASE1 and it took 150 seconds to populate DATABASE2.  Network was discarded as I recorded
in the package how long each of the 5 10000 chunk took to insert in each of the two databases as follows:
Records Being Inserted  Time it took in DATABASE1     Time it took in DATABASE2
First  10000             3 seconds                    27 seconds
Second 10000             4 seconds                    26 seconds
Third  10000             6 seconds                    40 seconds
Fourth 10000             4 seconds                    31 seconds
Fifth  10000             4 seconds                    26 seconds
When I look at the explain plan in both databases I see following:
| Id  | Operation                | Name | Cost  |
|   0 | INSERT STATEMENT         |      |     1 |
|   1 |  LOAD TABLE CONVENTIONAL |      |       |
My questions:
     1) Does the explain plan indicate that Direct Load was not used.
     2) If the answer to 1 is Yes, is it possible to use Direct Load or a faster insert method in this case?
     3) Any ideas what could be causing the 7.5 to 1 difference between the two databases.
Please note that these two databases are non production so load is negligible.
CREATE TABLE TABLE1
  TABLE1_ID                VARCHAR2(255)   NOT NULL,
  NUMBER_COLUMN_1          NUMBER,
  NUMBER_COLUMN_2          NUMBER,
  NUMBER_COLUMN_3          NUMBER
ALTER TABLE TABLE1 ADD CONSTRAINT TABLE1_PK PRIMARY KEY (TABLE1_ID);
CREATE INDEX NUMBER_COLUMN_3_IX ON TABLE1(NUMBER_COLUMN_3);
CREATE OR REPLACE PACKAGE TABLE1_PKG IS
  TYPE VARCHAR2_ARRAY      IS TABLE OF VARCHAR2(4000);
  TYPE NUMBER_ARRAY        IS TABLE OF NUMBER;
  TYPE DATE_ARRAY          IS TABLE OF DATE;
  PROCEDURE Insert_Table1
    Table1_Id_Array_In         TABLE1_PKG.VARCHAR2_ARRAY,
    Number_Column1_Array_In    TABLE1_PKG.NUMBER_ARRAY,
    Number_Column2_In          TABLE1_PKG.NUMBER_ARRAY,
    NUMBER_COLUMN_3_In         NUMBER
END;
CREATE OR REPLACE PACKAGE BODY TABLE1_PKG IS
  PROCEDURE Insert_Table1
    Table1_Id_Array_In         TABLE1_PKG.VARCHAR2_ARRAY,
    Number_Column1_Array_In    TABLE1_PKG.NUMBER_ARRAY,
    Number_Column2_In          TABLE1_PKG.NUMBER_ARRAY,
    NUMBER_COLUMN_3_In         NUMBER
  IS
  BEGIN
    FORALL I IN 1..Table1_Id_Array_In.Count
      INSERT /*+ APPEND */ INTO TABLE1 (TABLE1_ID            , NUMBER_COLUMN_1           , NUMBER_COLUMN_2     , NUMBER_COLUMN_3   )
      VALUES                           (Table1_Id_Array_In(I), Number_Column1_Array_In(I), Number_Column2_In(I), NUMBER_COLUMN_3_In);
  END Insert_Account_Keys;
END;
Thanks,
Thomas

Answer 1

I found answer for why Direct Path is not used when I do an insert into TABLE1@SOMEDATABASE SELECT....:
  http://docs.oracle.com/cd/E11882_01/server.112/e41084/statements_9014.htm#i2163698
Direct-path INSERT is subject to a number of restrictions. If any of these
restrictions is violated, then Oracle Database executes conventional INSERT serially
without returning any message, unless otherwise noted:
You can have multiple direct-path INSERT
Queries that access the same table, partition, or index are allowed before the direct-path INSERT
If any serial or parallel statement attempts to access a table that has already been modified by a direct-path INSERT
The target table cannot be of a cluster.
The target table cannot contain object type columns.
Direct-pathINSERT
Direct-pathINSERTAPPENDAPPEND_VALUESINSERT
The target table cannot have any triggers or referential integrity constraints defined on it.
The target table cannot be replicated.
A transaction containing a direct-path INSERT
My table is being replicated and I am try it via a distributed transaction.
I am still puzzled as to why it took 2 minutes and 44 seconds to insert 10000 rows in our production database but that's something that I'll investigate if time permits.  For now I've rewritten the process to use Insert...select if the number of records in the batch is less than or equal to a configured (currently set at 400000) number, else it will move using chunck and for now using bulk collect in the source pass arrays of data and FORALL inserts in the target.  If time allows in the future I will try to rewrite to use chunking combinde with insert..select.
Thanks to all for your help,
Thomas