Database Reference
In-Depth Information
longer-term analysis, auditing, and archival is stored in the Oracle database.
This is analogous to the general architecture depicted in Fig.
13.4
.Moreover,
the scenario above can be distributed. For example, the Northwind company
may have a centralized Oracle database and many applications running at
several application server nodes in various countries. To perform analysis of
orders and sales in near real time, we may install an IMDB cache database
at each node. On the contrary, the customer profiles do not need to be stored
at every node. When a node addresses a sales order, the customer's profile is
uploaded from the most up-to-date location, which could be either a node or
the central database. When the transaction is finished, the customer's profile
is updated and stored back into the central database. The IMDB cache can
also be used as a read-only cache, for example, to provide fast access to
auxiliary data structures, like lookup tables. On the other hand,
data aging
is an operation that removes data that are no longer needed. There are two
general types of data aging algorithms: the ones that remove old data based
on a timestamp value and the ones that remove the least recently used data.
Like the other systems commented in this chapter, TimesTen uses
compression
of tables at the column level. This mechanism provides
space reduction for tables by eliminating duplicate values within columns,
improving the performance of SQL queries that must perform full table scans.
Finally, TimesTen achieves
durability
in a similar way as SAP HANA,
that is, through transaction logging over a disk-based version of the database.
TimesTen maintains the disk-based version using a
checkpoint
operation
that occurs in the background, with low impact on database applications.
TimesTen also has a blocking checkpoint that does not require transaction
log files for recovery and must be initiated by the application. TimesTen uses
the transaction log to recover transactions under failure, undo transactions
that are rolled back, replicate changes to other TimesTen databases and/or
to Oracle tables, and enable applications to detect changes to tables.
In addition to the above, Oracle also commercializes an appliance called
Oracle Exalytics In-Memory Machine
,
7
similar to the SAP HANA
appliance studied in Sect.
13.5.4
. Exalytics is composed of hardware, business
intelligence software, and an Oracle TimesTen IMDBS. The hardware consists
in a single server configured for in-memory analytics of business intelligence
workloads. Exalytics comes with the Oracle BI Foundation Suite and Essbase,
a multidimensional OLAP server enhanced with a more powerful MDX syntax
and a high-performance MDX query engine. Oracle Exalytics complements
the Oracle Exadata Database Machine, which supports high performance for
both OLAP and OLTP applications.
Search WWH ::
Custom Search