Performance Optimization

During an initial load, when does SLT ALWAYS use reading type 1, regardless of what you configured?

Which reading type does SLT default to when you pick the Resource Optimized initial load mode (non-HANA source)?

Reading types 4 and 5 behave almost identically. What single property decides which one you must use for a given table?

For generic table ranges in replication, how many ranges should you define relative to the number of parallel jobs you want?

Enabling generic table ranges adds a field to the logging table and stores a hash. What is the field name and the hash range?

For generic ranges, what makes a key field a poor grouping-field choice, and what's the risk of piling on more key fields to compensate?

Generic table ranges (Use Key Fields to Calculate Ranges) aren't available on every DB. Which table class is also excluded even on supported databases?

You're already using Specify Ranges Manually for a replicating table and want to change the number of ranges. What must you do first?

For the two logging-table processing options in LTRC, which is the default and when should you override it?

With Only Process Logging Tables with Records and a SECONDARY DATABASE connection, when are source structure changes detected?

Starting with DMIS 2020 / S/4HANA 1909, subscription-based change data capture is used. Which reading type and table types does it support?

You choose Performance Optimized initial load mode on a non-HANA source. Which reading types will SLT use, and what's the prerequisite you must confirm first?

You accidentally assign reading type 5 to a cluster table (or reading type 4 to a transparent/pool table). What happens?

Reading type 3 has a hard restriction on filtering that the other reading types don't. What is it?

In reading type 1 (range calculation), what makes a field a GOOD choice as the selection parameter for splitting a table?

For reading type 1 on a non-HANA source, what indexing requirement must the range/selection field satisfy?

When you manually define ranges in SLT (reading type 1 or 4/5), what is the one rule you must never violate?

Above what table size does the book recommend segmenting a table into ranges to speed up initial-load portion calculation?

What is the default package (portion) size for an SLT initial load, and what is the maximum you can set?

How is the package size for an initial-load portion actually computed?

What is the default read-portion size for reading type 3, and what unit is it measured in?

For reading type 3, how much data can a single access plan hold, and how many portions does that translate to by default?

Which backend table temporarily holds the compressed sender-queue data for reading types 4 and 5, and what happens to it after the load?

How much extra source-system storage do reading types 4/5 require for a CLUSTER table versus a POOLED table?

By default, how many jobs does SLT allocate to load a single table in parallel, and where do you change it before the load starts?

A table is already loading and you need to increase its parallelism NOW. Which transaction do you use, and why not the usual one?

When access plans are computed on the primary key, why does the book cap concurrent calculation jobs per table at 8?

The Number of Records parameter (access-plan splitting) has both a hard minimum and a recommended minimum. What are they?

You set Number of Records to 40 million on a 225-million-record table. How many access plans does SLT generate, and how are the records split?

For a cluster table like BSEG, which record count must you use when sizing the Number of Records access-plan parameter?

For sizing initial-load jobs, roughly how many records per hour can one job move for a transparent table versus a cluster table?

During ongoing REPLICATION (not initial load), what per-second change rate suggests a table would benefit from optimization?

A colleague wants to speed up an initial load and configures filter RULES to cut the data volume. Why won't this help performance?

Between the three initial-load tuning levers (reading type, parallel jobs, filtering), which does the book call the MOST effective, and why?