Submitting forms on the support site are temporary unavailable for schedule maintenance. If you need immediate assistance please contact technical support. We apologize for the inconvenience.
Determining the amount of disk space needed when using Shareplex
설명
How do I determine the amount of disk space needed when operating with Shareplex?
해결 방안
Disk space for queue files - Queue File Size:
The main SharePlex consumers of disk space are the queues, which are periodically written from memory to disk in the rim sub-directory within the SharePlex variable-data directory. To accommodate the data residing in a queue (capture, export, post), SharePlex builds data files for each queue on demand. All queue data files are the same size, the default size being 8 MB. Since the queues are dynamic, the number of data files can grow considerably when replication is stopped, either intentionally by a user or in the case of a network instance or system failure. If the network or target system is down or unavailable, data is queued in the export queue on the source system. If the target instance is down, data is queued in the post queue on the target system. As a result, you need enough disk space on both source and target systems to hold the queue data for as long as the downtime persists.
To determine the amount of space you need, use this formula:
1. Use a worst-case scenario and estimate the length of downtime to be tolerated. 2. Determine how much data is replicated in that time, using the below formula as a guide.
This formula multiplies the redo log value by 1/3 because only about 1/3 of the redo log is actually data that is replicated. The rest is data that Oracle uses in maintenance and operation of the instance itself.
[size of a redo log] x [number of log switches in 1 hour] x 1/3 x [number of hours downtime] = amount of disk space needed for the queues on each system
For example, if you expect to recover from 8 hours of downtime, and your redo logs are 60 MB, and they switch five times an hour, then you could need 800 MB of space on both source and target machines for the SharePlex queues:
[60 MB redo log] x [5 switches/hour] x [1/3] x [8 hours] = 800 MB disk space
