Databases Reference
In-Depth Information
people employ a so-called
CRUD matrix
that contains Create, Read, Update, and
Delete indicators, or even estimates for how many operations will be performed for
each key object used by a business transaction. These estimates may be per minute,
per hour, per day, or for whatever time period makes sense in the context of your
system. For example, the CRUD matrix for a simple employee update transaction
might be as shown in
Table 3-1
, with the checkmarks indicating that each transac‐
tion performs the operation against the object shown.
Table 3-1. Access patterns for database objects
Object
Create
Read
Update
Delete
EMP ✓ ✓
DEPT
✓
SALARY
✓ ✓
How much hardware do I have now, and how much will I add as the database grows?
Disk drives tend to get cheaper and cheaper. Suppose you're planning a database of
10 TB that you expect to grow to 90 TB over the next two years. You may have all
the disk space available to plan for the 90 TB target, but it's more likely that you'll
buy a smaller amount to get started and add disks as the database grows. It's im‐
portant that you plan the initial layout with the expected growth in mind, although
Real Application Clusters make it much easier to scale out your database easily.
Prior to Oracle9
i
, running out of tablespace in the middle of a batch operation
meant that the entire operation had to be rolled back. Oracle9
i
introduced the
concept of
resumable space allocation
. When an operation encounters an out-of-
space condition, if the resumable statement option has been enabled for the session,
the operation is suspended for a specific length of time, which allows the operator
to correct the out-of-space condition. You even have the option to create an AFTER
SUSPEND trigger to fire when an operation has been suspended.
With Automatic Storage Management (ASM), introduced in Oracle Database 10
g
,
you can add additional disk space or take away disks without interrupting database
service. Although you should still carefully estimate storage requirements, the pen‐
alty for an incorrect judgment, in terms of database downtime, is significantly re‐
duced with ASM.
What are the availability requirements?
What elements of redundancy, such as additional disk drives, do you need to pro‐
vide the required availability? ASM also provides automatic mirroring for data with
different redundancies available, which can help to provide data resiliency.
