# GSLB decision-flow explainer > The BIG-IP DNS two-tier decision, explained: pool selection at the wide IP, then the preferred, alternate and fallback chain inside the pool, with the grammar validated and the manual's rules cross-checked. - Tool: https://ronutz.com/en/tools/f5-gslb-decision-flow - Family: Networking --- # GSLB decision-flow explainer Paste `gtm wideip` and `gtm pool` stanzas (as `tmsh list gtm wideip a ` and `tmsh list gtm pool a ` print them) and the two-tier BIG-IP DNS (formerly GTM - Global Traffic Manager) decision renders as it really runs. The wide-IP tier shows the pool-selection method (`pool-lb-mode`, default round-robin) with the attached pools, their order and ratio values, and the wide-IP settings that shape answers: persistence, last-resort-pool, failure-rcode behavior, and the decision-log verbosity knob that lets you watch this same flow in the logs. The pool tier renders the three-step chain: preferred (`load-balancing-mode`), alternate (`alternate-mode`), fallback (`fallback-mode`), each step carrying the vendor's semantics for the tier and the method. The defaults are applied and labeled when an attribute is absent: round-robin, round-robin, return-to-dns. Underneath, deterministic observations cross-check the configuration against the documented grammar and the Load Balancing manual's rules: methods outside a tier's admitted token list, the fallback tier's ignores-availability property stated on every resolved chain, Fallback IP wiring (the method with no address, or an address no tier uses), the dynamic-ratio applicability set, zeroed QoS coefficients, member ratios no tier consumes, the Global Availability pairing rule, and the topology-at-both-tiers warning that wants each pool's fallback set to None. A single method name explains that one method, including which tiers admit it and where its decision signal comes from (LDNS-path probing, server-side statistics, or none). The word `methods` lists both catalogues. Everything runs locally; nothing you paste leaves the page. ## Standards and references - [F5 TMSH Reference: gtm pool a (load-balancing-mode, alternate-mode, fallback-mode, dynamic-ratio, qos-* coefficients)](https://clouddocs.f5.com/cli/tmsh-reference/v15/modules/gtm/gtm_pool_a.html) - the three-tier mode grammars with per-token descriptions, the chain defaults, the dynamic-ratio applicability set, and the pool options the observations cross-check - [F5 TMSH Reference: gtm wideip a (pool-lb-mode, persistence, last-resort-pool, decision-log verbosity)](https://clouddocs.f5.com/cli/tmsh-reference/latest/modules/gtm/gtm_wideip_a.html) - the wide-IP pool-selection grammar and semantics, including the v13 reference's version-dependent random token - [BIG-IP DNS: Load Balancing - About load balancing and resource availability (and the Topology chapter's both-tiers warning)](https://techdocs.f5.com/en-us/bigip-15-0-0/big-ip-dns-load-balancing/about-load-balancing-and-resource-availability.html) - the chain semantics: alternate is static-only, fallback ignores availability, the None cascade and the BIND aggregate, the dynamic-ratio proportional example, and the topology-at-both-tiers fallback rule ## Related reading - [BIG-IP DNS Load Balancing: the Wide IP, the Pool, and the Three-Step Chain](https://ronutz.com/en/learn/gtm-load-balancing-methods.md): A GSLB answer is decided twice: the wide IP picks a pool, then the pool picks a member through a preferred, alternate and fallback chain. The chain carries the rules people trip over: the alternate can only be static, the fallback ignores availability on purpose, and None cascades all the way to a BIND aggregate. - [GTM Topology Records: Longest Match Is a Sort, Not the Pick](https://ronutz.com/en/learn/gtm-topology-records-and-longest-match.md): Topology load balancing scores candidates from an ordered record list, and the ordering is what Longest Match actually does. The first record to match a candidate scores it, later records are shadowed, and the highest score wins, which is why a heavy wildcard really can beat a light /32.