How to Optimize Performance of SAP NW MDM System

[Former Member]

Introduction

This document is compilation of setting and customization that can be done in order to improve the performance of your SAP NW MDM System. Every Parameter has definition with Recommended Value.

MDIS Global Parameter

Interval

The number of seconds MDIS waits after scanning ports before restarting scanning.

Recommended value: 300

Tracing Level

Tracing Level lets you customize the minimum severity level recorded to meet your own requirements

Available values:

a)debug

b)flow

c)info

d)warning

e)error

Recommended Value: Set it to Warning as it has least impact on performance of server.

MDIS Parameters Repository Level

Parameter "Inter-Chunk Delay MS":

If data imports are performed during online working hours then users cannot access the repository due to long-running imports, In order to resolve this problem set this parameter to a reasonable value to allow for repository access between imports of chunks. This parameter can be set for individual repository.

Recommended value: 5000

File Aggregation

In certain scenarios, it may be more efficient for MDIS to import files from a port in batches rather than as individual files. For example, if you are generating large numbers of small import files (containing one or two records each), MDIS can import these files faster by processing them collectively than it could by processing each file separately

Recommended value: 10-50

Chunk Size

The Chunk Size parameter defines the number of virtual extended records to include in a chunk. Typically, the smaller the chunk size the better, as smaller chunks can be processed faster within MDIS and also require less wait time in transactions between MDIS and the Master Data Server

Recommended value: 50000

No. of Chunks Processed In Parallel

The No. of Chunks Processed in Parallel parameter optimizes the way chunks are processed within MDIS. This parameter determines how many chunks MDIS can manage simultaneously.

Recommended value: 5-10

MDSS Global Parameter

Tracing Level

Tracing Level lets you customize the minimum severity level recorded to meet your own requirements

Available values:

a)debug

b)flow

c)info

d)warning

e)error

Recommended Value is Warning as it has least impact on performance of server.

Suppress unchanged records option

You can limit the records included in a syndication file to only those that have changed since the last time they were syndicated to the mapped remote system. This improves the performance of syndication server.

Recommendation: If your business requirements allow it, enable the option that suppresses unchanged records in the MAP properties in order to reduce the number of records involved in syndication and reduce job run-times. Record suppression can be done at remote system level or port level

Suppress records without key option

Recommendation: If your business requirements allow it, enable the 'Suppress Records Without Key' option in the MAP properties to reduce the number of records involved in syndication and reduce job run-times.

Import and Syndication schedule without overlaps

In case of multiple repositories in single server multiple repositories, you should be aware that imports and syndication are functions that are scheduled sequentially by the MDM Server. If many of these tasks are started at the same time, the processes are handled sequentially. This can cause performance bottlenecks.

Recommendation:

Imports should be scheduled so they are not performed during user operation time. Syndication also should not run during import processes. Try to schedule imports so that they do not overlap with syndication.

No parallel imports into multiple repositories

The import process will not only lock the repositories, but it also has to place short server locks on the MDM Server. For this reason, performance losses can be expected if you import data to several repositories at the same time.

Recommendation: In case of multiple repositories,   put together a schedule so that you do not have to import data to several repositories at the same time.

Repository Designing Considerations:

Sort Index:

Sort Index creates an in-memory sort index for the field and enables sorting on the UI in ascending or descending order.

Setting "Sort Index" to "Normal" for a large number of fields can have a negative impact on the repository loading time and main memory consumption.

Recommendation: Make only those fields sortable for which are required for matching and sorting.

Workflow Designing:

Workflows provide a powerful way to automate different processes within MDM System.

Recommendation: Avoid workflow loops as it degrades system performance. For workflow which are completed either delete them at end or archive them if report is necessary for audit purpose.