Configuring staging space resources on z/OS for the incremental update process
This document describes how to configure the staging space, which determines whether InfoSphere Data Capture (CDC) can handle a particular workload. It contains instructions on how to define a suitable configuration and gives advice with regard to events and counters worth monitoring during that process. Finally, the text discusses the impact that the configuration settings have on the CDC Capture Agent.
The incremental update function of the IBM DB2 Analytics Accelerator for z/OS tracks and stages all transactions in the system memory until they are committed (and subsequently transferred to the accelerator) or rolled back. The component responsible for this task is the CDC Capture Agent on z/OS. Sufficient memory must be allocated to this component. If the memory does not suffice, a staged transaction might stop incremental updates for all connected accelerators. The error message CHC0330E is issued in such cases.
Configuration settings relevant to the staging space
The CDC configuration parameter MAXSUBSCRSTAGESIZE defines the amount of memory for the staging of incremental update transactions to be processed by an accelerator. So the total amount of memory that must be available to the CDC Capture Agent is roughly this amount multiplied by the number of accelerators. The total amount is configured by the configuration parameter STG64LIMIT. However, the required memory also depends on the actual workload. The default values are therefore only starting points for an evaluation. The following website provides heuristics, which might serve as a computational basis:
Testing settings relevant to the staging space
Because the staging space size depends on the workload, it is crucial that you test the configuration settings with the anticipated workload. To make sure that sufficient staging space is allocated, monitor the following events in the log of the CDC Capture Agent:
|Event||Sketch of semantics|
|CHC0051W||The oldest open commit group in the staging space has exceeded a certain threshold. Long running transactions can occupy a significant amount of the staging space for a longer period of time and thus may lead to the space reaching its overall limits.|
|CHC9618W||Indicates that the overall memory consumption has reached a defined limit|
|CHC1532W||A subscription has occupied more than 90% of the available staging space|
|CHC1524E||A subscription has used up all of the available staging space (100%)|
|CHC0330E||Indicates that the staging space is full and that the system has to abort.|
For high workloads, it might prove useful to create staging space reports. These reports show the content of the staging space and other details, which might help you understand where the problems are:
Example: F <cdcjob>,REPORT,TYPE=STGSP,NAME=<subname>
As always when monitoring the memory consumption, it is helpful to monitor the storage manager:
Background information: impact of configuration settings on the CDC Capture Agent
The CDC Capture Agent uses a configurable amount of above-the-bar storage (STG64LIMIT) for the staging of transactions and for the so-called retry cache. If this additional space is not configured, the agent uses its regular amount of storage, which is no more than 2GB because the product is a 31-bit application. If the above-the-bar storage is exhausted, the component will try to keep going with the regular storage. If the regular storage is also exhausted, memory allocation (malloc) failures occur. These memory allocation failures have a varying impact depending on the part of the product in which they occur.
|Information Management||DB2 Analytics Accelerator for z/OS|
Translate this page: