AWS · Infrastruttura · YoctoIT tech page

Elasticità

The cloud's true elasticity: managed load balancers and capacity that follows the load — multi-AZ availability as a setting, not a project.

FOCUS · HIGH AVAILABILITYMulti-AZ by design: the application that survives the failure of an entire data center
Request the Free Assessment Learn more
YoctoIT material for clients and partners · AWS e i nomi dei servizi sono marchi di Amazon.com, Inc.
01 · What it is

Elastic Load Balancing & Auto Scaling, made clear.

ALB for modern HTTP traffic, NLB for heavy TCP, Gateway LB for appliances: AWS load balancers are managed, scale by themselves and distribute across multiple Availability Zones. With Auto Scaling Groups, instances are born and die following the metrics.

3 AZ
the bare minimum of our architectures: a zone failure is not an incident
Health check
traffic goes only to healthy targets: failover is automatic
Zero pre-provisioning
capacity follows the load: no iron sized for the peak
Icona ufficiale AWS — Elastic Load Balancing & Auto Scaling
OFFICIAL AWS ICON · Elasticità
02 · How to use it well

The things that make the difference.

The traffic flow

Users / applicationsthe incoming load
ALB · HTTP/S
NLB · TCP/UDP
Gateway LB
the right load balancer per protocol
Multi-AZ target groupsinstances distributed across zones
Auto Scaling Groupscale-out on peaks, scale-in at night
From request to instance, without single points

Application Load Balancer

Path and host routing, managed TLS, WAF integrable: the front end of web applications.

Network Load Balancer

Millions of connections at minimal latency: the load balancer for serious TCP workloads.

Scaling policies

Target tracking on the right metrics: CPU isn't always the one that matters.

Warm pools & lifecycle

Pre-warmed instances and lifecycle hooks: scale-out ready even for slow boots.

03 · In depth

ALB, NLB and scaling that anticipates

ALB works at L7 (host/path routing, WebSocket, gRPC, integrated OIDC), NLB at L4 with millions of connections and static IPs, Gateway LB inserts security appliances inline. The Auto Scaling group combines health checks, warm pools and policies: target tracking (keep CPU at 60%), step and predictive scaling that learns the patterns. The secret is the right grace period and immutable AMIs to scale in seconds.

  • ALB L7 — host/path/header routing, OIDC authentication in front of the apps
  • NLB L4 — static IPs, TLS passthrough and millions of connections: the raw load
  • Target tracking — 'keep CPU at 60%': the policy that tunes itself
  • Predictive scaling — ML on historical patterns: capacity ready BEFORE the peak
  • Warm pool — pre-initialized instances: scale-out in seconds, not minutes
  • Health check — ELB + EC2 combined: the sick node leaves before doing damage
04 · Numbers and lifecycle

The numbers that matter.

L4/L7
the two planes: NLB for raw traffic, ALB for the application layer
60 s
the typical scale-out with immutable AMIs and warm pools
0
downtime in deploys with rolling + connection draining
3 AZ
the standard for target distribution
Elasticity is discipline: immutable images, honest health checks and policies tuned on YOUR patterns — we configure it and put it to the test.
05 · Use cases

Where it really pays off.

E-commerce under peak

Black Friday without sizing the year on one day: elastic capacity and an honest bill.

Mission-critical APIs

The backend that knows no maintenance windows: rolling updates behind the load balancer.

Nightly batches

Fleets born at 2 and gone by 5: the work done, the cost zeroed.

We design multi-AZ by default: a zone failure is a non-event — and we prove it in failover tests.