System Architecture & Instances

The enqueue server holds its lock table in the memory of the central services instance — what does that storage choice imply if that instance restarts without protection?

You suspect ONE specific dialog instance (not the one you're logged into) is saturating its work processes — which of SM50 / SM66 / SM51 gets you there fastest?

How many message servers exist in a system, and why does that count make logon load balancing a single point of dependency?

A colleague insists the primary application server (PAS) must never be taken offline because it is functionally special — are they right?

If ABAP loads live in the database (not on each app server's disk), what does that explain about a freshly started instance's first-access performance?

A teammate wants to bounce all three landscape systems together because they 'share infrastructure' — what's wrong with that mental model?

Why is it acceptable to skimp on backing up an application server but never the database?

Symptoms: (a) one user's session dies and a process restarts, vs (b) the entire instance goes down at once — which points to a work process crash and which to a dispatcher crash?

What is the ABAP platform's relationship to SAP NetWeaver and to S/4HANA?

A work process is consuming huge heap memory and then gets killed and restarted — what is the chain of events, and why heap rather than the normal pool?

The SAP Web Dispatcher load-balances HTTP(S) just like the message server balances dialog logons — so why does it get central-services-grade HA, not app-server-grade?

What is the valid range for an instance number, and which values are reserved?

Why is the production recommendation 'at least two dialog instances' framed as availability rather than the cluster/HA protection you give central services?

The message server has failed. Logged-on users are still working fine, but no one new can log in — does this rule the message server in or out as the cause?

To make the central services instance highly available, why can't you just run a second one the way you'd add a second dialog instance?

Inbound HTTP/HTTPS (and SMTP) to an instance — is that the gateway's job or the ICM's, and where does it hand off?

What five types of work process can run on a dialog instance?

What are the minimum work-process counts — and which are per dialog instance vs system-wide?

For a production system, which hardware configuration must you NOT use, and where's the line between acceptable and not?

HANA runs only on Linux (SLES/RHEL). In a two-level configuration, what does that force about your application servers — and why isn't that the production choice anyway?

What are the rules for a System ID (SID), and which names are reserved and cannot be used?

After a central services failover, locks would normally be lost — what does the enqueue replication server (ERS) change, and what does it NOT replace?

Users can't log on, but the dialog instances are up and only central services is under maintenance — should logons be affected?

An instance has two names — when you're editing parameter profiles, which form do you need, name 1 or name 2?

An RFC connection from a non-SAP system to one of your dialog instances fails — which per-instance process should you suspect first?

Why does putting two instances with the same instance number on one server break things even if they belong to different SAP systems?

Why are the dispatcher and every type of work process the same executable (disp+work), and what does that shared-binary fact imply?

In a hybrid landscape you must connect on-premise SAP S/4HANA to SAP BTP securely. Where is the Cloud Connector placed, and which side establishes the connection?

You're setting up the Cloud Connector for production use. Where do you get it, which package variant should you pick, and how do you reach its admin UI?

Cloud Connector SAP BTP subaccount certificates need periodic renewal. Has that always required manual work, or can it be automated?