The following table identifies the recommended data-type mappings for Database Ingestion and Replication configurations with a Oracle source and a Databricks target:
Oracle Source Data Type
Databricks Target Data Type
binary_double
double
binary_float
float
blob
binary
char(s byte), 1 <= size <= 2000
string
char(s char), 1 <= size <= 2000
string
clob
string
date
timestamp
float(precision), 1 <= p <= 126
string
integer
string
long raw
binary
long(2147483648 byte)
string
nchar(s char), 1 <= size <= 2000
string
nclob
string
number
string
number(*,s), -84 <= s <= 127
string
number(p,s), 1 <= p <= 38, -37 <= s <= 38
decimal(p,s), 1 <= p <= 38, 0 <= s <= 38
number(p,s), 1 <= p <= 38, -84 <= s <= 127
string
nvarchar2(s char), 1 <= size <= 4000
string
raw(size), 1 <= size <= 2000
binary
rowid
string
timestamp(precision) with time zone, 0 <= p <= 6
timestamp
timestamp(precision) with time zone, 7 <= p <= 9
string
timestamp(precision), 1 <= p <= 6
timestamp
timestamp(precision), 7 <= p <= 9
string
varchar2(s byte), 1 <= size <= 4000
string
varchar2(s char), 1 <= size <= 4000
string
xml
string
LOB limitations
Database ingestion and replication initial load, incremental load, and combined load jobs can replicate data from Oracle BLOB, CLOB, LONG, LONG RAW, NCLOB, and XML columns to Databricks targets. LOB column data might be truncated on the target. For all of the Oracle LOB data types replicated to this target type, the truncation point is 16777216 bytes. For more information, see the "Include LOBs" description in Configure an Oracle source.
Unsupported source data types
Database Ingestion and Replication does not support the following Oracle source data types:
•"ANY" types such as ANYTYPE, ANYDATA, ANYDATASET
•Extended types
•INTERVAL
•JSON
•BFILE type for storing LOB data externally
•NUMBER with a scale greater than the precision
•UROWID
•Spatial types such as SDO_GEOMETRY
•User-defined types such as OBJECT, REF, VARRAY, nested table types
Source columns that have unsupported data types are excluded from the target definition.
If necessary, you can create custom data-type mappings when you define a database ingestion and replication task. For more information, see "Data Type Rules" in Configure a Databricks target.