Exercise 2: Five LoadBalancer Services vs. One Ingress — Possible Solution
====================================================================

COST ARGUMENT:

Per the chapter, "a LoadBalancer Service provisions ONE external cloud
load balancer PER Service." Using five separate LoadBalancer Services
for the five microservices means provisioning FIVE separate cloud load
balancers, each carrying its own ongoing cost (directly tying to
`cloud1-9`'s cost management material) and its own separate external
IP address to manage. A single Ingress, by contrast, uses ONE shared
external entry point (typically backed by just one load balancer,
provisioned for the Ingress controller itself) to route to all five
backend services -- reducing this from five billed load-balancer
resources down to effectively one, a real, direct cost saving that
scales with however many services need to be exposed this way.

LAYER 4 VS. LAYER 7 CAPABILITY ARGUMENT:

The scenario specifically requires all five services to be reachable
"under the same domain, on different paths" -- this is fundamentally
a PATH-BASED ROUTING requirement. Per the chapter, a plain LoadBalancer
Service "operates at Layer 4... it has no understanding of HTTP paths
or hostnames at all, just raw TCP/UDP forwarding." Five separate
LoadBalancer Services would each need their OWN separate external IP
or hostname -- there would be no way to have all five sensibly share
ONE domain with different paths routing to different services, since
Layer 4 forwarding has no concept of a URL path at all.

An Ingress, by contrast, operates at Layer 7 specifically to solve
this -- per the chapter, Ingress rules can express exactly "example.com/
api/* routes to api-service, example.com/other/* routes to
other-service" and so on for all five, all sharing the SAME domain and
external entry point, which is precisely the capability a plain
LoadBalancer Service structurally cannot provide at all, regardless of
cost.

Combined conclusion: a single Ingress is the better choice both because
it's meaningfully cheaper (one load balancer instead of five) AND
because it's the only option of the two that can actually satisfy the
stated requirement (path-based routing under one shared domain) at
all -- five LoadBalancer Services couldn't fulfill this requirement
even ignoring cost entirely.

WHY THIS WORKS AS AN ANSWER
------------------------------
This addresses both dimensions the exercise explicitly asks for --
cost and Layer 4/7 capability -- as two SEPARATE, independently
sufficient reasons Ingress is the better choice here, directly citing
the chapter's own stated limitations of LoadBalancer Services for
each.