Performance Optimization

In reading type 1, what happens if you leave the Selection Parameter field blank?

In the Specify Transfer Settings step, what initial load mode is the default and which reading type does it apply to all tables?

When you pick Performance Optimized as the initial load mode, which reading types get applied to transparent versus cluster tables?

In an ODP scenario (SLT feeding SAP Data Services), how do you start and stop table replication?

After an SLT outage, why can replication-phase tables resume where they left off while initial-load/calculation tables cannot?

Starting with DMIS 2020 / S/4HANA 1909, subscription-based change data capture is available. What are its two hard constraints?

What are the boundaries where reading type 3 stops being appropriate?

A small table's calculation phase runs in a single process even though you configured parallel jobs. Is something wrong?

A logging table exists in the source for a replicated table. What triggers an entry to be written into it?

For reading types 1 and 3 versus 4 and 5, what do the No. of Calculation Jobs actually do?

The source system runs on an SAP HANA database. Which reading type will SLT use for the initial load regardless of the load mode you chose?

You specify reading type 5 for a cluster table by mistake (it should be type 4). What does SLT do at design time and at runtime?

What is the one property that decides whether a table uses reading type 4 versus reading type 5?

Reading type 1 is described as one of the earliest methods. In today's SLT, when is it actually used?

For reading type 1, why is a field like CLIENT a poor selection parameter, and what makes a good one?

When defining ranges for reading type 1 on a huge table, what single mistake causes data to be loaded twice?

How is the package size for reading type 1 computed, and what is the maximum you can specify?

Reading type 3 is the default. What lets it work without any additional index on the source table?

What is the default portion size for reading type 3, and where do you change it?

How many records can a single access plan process, and how many portions does that translate to?

You need a filter on an initial load but the table uses reading type 3. What blocks you?

How do reading types 4 and 5 actually extract data before it reaches SLT?

How much extra source-system storage do reading types 4 and 5 need for cluster tables versus pooled tables?

Which table holds the sender-queue data for reading types 4/5, and what is its lifecycle during a load?

For a non-ABAP source using reading type 4 or 5, where does the temporary table DMC_INDXCL live?

What is the default portion size for reading types 4 and 5, and where do you change it?

For a very large table under reading type 4/5, why is raising the package size the wrong move, and what do you do instead?

A table has 225 million records and you set the Number of Records per access plan to 40 million. How many access plans run, and how are records split?

When setting Number of Records for a cluster table, which record count do you base it on?

What is the hard minimum and the recommended minimum for the Number of Records per access plan?

When access plans are computed on the primary key, why cap concurrent jobs per table at eight?

For an initial load of a table over a billion records, how do you split it into ranges, and where are the range entries stored?

By default, how many parallel jobs does SLT allocate to load one table, and where is this changed?

You need to change the parallelism of an initial-load job that is already running. Which transaction do you use?

Filter rules and filter conditions both reduce data, but only one improves load performance. Which, and why?

You want to filter for a numeric-only alphanumeric field to keep values greater than 10,000 in LTRS. What is the entry gotcha?

As a planning rule of thumb, how many records can one background job load per hour for a transparent table versus a cluster table?

During ongoing replication, what change rate signals a table might benefit from replication-side optimization?

What does the default Only Process Logging Tables with Records option do, and how does Process All Logging Tables differ?

You have 100+ replicated tables. When should you choose Process All Logging Tables over Only Process Logging Tables with Records?

Under Only Process Logging Tables with Records over a secondary database connection, when are source-table structural changes detected?

To speed up replication of one busy table, you divide its logging table into ranges. What field does the generic approach add, and what does it store?

When must generic table ranges (Use Key Fields to Calculate Ranges) be configured, and what if you need them for a table already replicating?

The Use Key Fields to Calculate Ranges option is greyed out for your source. What does that mean, and which table class also can't use it?

When dividing a logging table into ranges for parallel replication, how many ranges should you define relative to the number of parallel jobs you want?

You want to change the number of ranges for a table that is currently replicating. When is it safe, and when must you reset the object?

Which temporary table holds the sender queue for subscription-based CDC, and how much extra storage does it need?

Among reading types, parallel jobs, and filtering, which lever gives the biggest initial-load performance gain, and why?

On a non-HANA source, what must exist on the field you use to define reading type 1 ranges?