Achieving Extreme Performance with Oracle Exadata in Software

Drawing QR Code JIS X 0510 in Software Achieving Extreme Performance with Oracle Exadata

decompression, which increases the amount of data compression while reducing the overhead needed to decompress the data, but also compresses data added through database writes without the overhead of compression spawned by every database change NOTE The OLTP designation in the name of this feature is somewhat misleading, as Advanced Compression is also used by many data warehouses, since compression is implemented for data added to the warehouse by many types of refresh operations that operate with smaller amounts of data Exadata Hybrid Columnar Compression is an Exadata feature that offers the benefits of compression in a slightly different way

Exadata Hybrid Columnar Compression (EHCC) is a type of compression that is only available on the Exadata Storage Server As its name implies, one of the distinguishing features of Exadata Hybrid Columnar Compression is that data using this type of compression is organized for storage by columns But there is much more to Exadata Hybrid Columnar Compression than simply a different type of storage The hybrid portion of the feature name refers to a difference in how Oracle stores columnar data, which allows better utilization across the complete spectrum of database operations, allowing for broader usage of Exadata Hybrid Columnar Compression In addition, Exadata Hybrid Columnar Compression uses different types of compression algorithm

Columnar storage is not a brand-new concept in the database world Other databases have offered columnar storage for a while, but universal columnar storage, even for a single table, can present problems Virtually no feature provides equal benefits for all types of operations, whether in a database or any other type of software The sweet spot for columnar storage is those tables where only a few columns will be accessed

3:

in a typical query In order for a database to satisfy a data request, the columns requested have to be accessed and assembled A larger number of columns results in more work for a pure columnar storage database when many of these columns are requested by a query And, unfortunately, queries that require a large number of columns coupled with a small number of rows are not that uncommon in the overall mix of a production database These queries end up requiring pure columnar databases to perform a lot of work for a small result set, reducing the benefit produced by columnar storage in general Exadata Hybrid Columnar Compression takes a slightly different approach, as shown in Figure 3-3 Exadata Hybrid Columnar Compression uses the concept of a compression unit, which stores data by columns, but only for a set of rows A compression unit can be retrieved with a single I/O operation, which is more efficient than a purely columnar storage option, which requires one or more I/O operations per column This advantage, in turn, means less unnecessary overhead for those queries that require a greater number of columns, which means you can use the advantages of Exadata Hybrid Columnar Compression for a broader set of tables, increasing the overall benefits you can gain from this feature Exadata Hybrid Columnar Compression creates compression units for data as the data is initially loaded into the database with direct load operations You can use an ALTER TABLE statement to modify how new direct loaded data is compressed, but the only way to get uncompressed data to use EHCC is to use the DBMS_REDEFINITION package to change the compression on a partition or table Data added to a table with Exadata Hybrid Columnar Compression with INSERT operations after this initial load is still compressed, but at a lower level of compression

FIGURE 3-3

The same conditions apply when data in a compression unit is updated the new data is removed from the compression unit and compressed with an alternate compression method, and the entire compression unit is locked for the duration of the update, which means that you typically only want to use Exadata Hybrid Columnar Compression for data that is no longer being updated As you look at the diagram of a compression unit, you can see that the overall unit holds data from all the columns in the table As data is loaded into a table or partition with Exadata Hybrid Columnar Compression enabled, the Exadata Storage Server Software estimates the number of rows that can fit into a compression unit, based on the size of the rows and the estimated compression of the data, using the Exadata Hybrid Columnar Compression compression algorithms