A production load balancer, minus the ops
Everything documented, everything managed from one console — health, routing, certs, logs.
Smart routing
Round Robin for even distribution, Source IP Hash for session stickiness, Least Connection for uneven request costs. Pick per backend group.
HTTP health checks
ALBs probe backends over HTTP — 2xx/3xx means UP, everything else is pulled from rotation until it recovers. Server health is visible live in the Stats tab.
SSL termination
Import your certificate (with private key and chain) into the SSL Certificate Manager, attach it to an HTTPS frontend, enable HTTP→HTTPS redirect, and optionally re-encrypt to backends.
Layer-7 ACL rules
Route by path, host, query parameter, HTTP method or source IP with exact/prefix/suffix/regex matching — first-match ordering with drag-to-reorder.
Auto-scaling backends
Point an ALB backend group at an Auto Scale Group and the LB lists and de-lists nodes as the group scales — no manual server edits.
Deep observability
Live HAProxy frontend/backend stats, Prometheus metrics, CPU/memory/disk monitoring, alerts, access logs to object storage, and a full activity timeline.
Application LB or Network LB?
Both share the same create flow — plan, target mapping, VPC, security group — but solve different problems.
| Capability | Application Load Balancer (ALB) | Network Load Balancer (NLB) |
|---|---|---|
| Layer | Layer 7 — HTTP / HTTPS / Both | Layer 4 — TCP |
| SSL termination | Yes, with imported certificates + HTTP→HTTPS redirect | No — TCP passthrough |
| Health checks | HTTP checks (2xx/3xx = UP) | TCP connect-level; HTTP checks are ALB-only |
| ACL content routing | Path, host, query, method, source IP | Not available with TCP |
| Auto Scale Group backends | Yes, with Scaler port | Node backends only |
| Routing policies | Round Robin · Source IP Hash · Least Connection | Same three; defaults to Least Connection |
| Listening ports | 80 / 443 per frontend protocol | 80, 443, 1024–65535 (excl. 8080, 10050, 9101) |
| Scheme | Internal or external | Internal or external |
| Typical use | Websites, APIs, microservices, WebSockets | Databases, MQTT, game servers, custom TCP protocols |
Plans are sized by vCPU, RAM, storage and max concurrent connections. Hourly and committed billing available.
How it works
From zero to balanced traffic in four steps.
Pick a plan & type
Choose a plan sized for your concurrent connections, then ALB (HTTP/HTTPS) or NLB (TCP), internal or external scheme.
Map your backends
Create backend groups, add servers by IP and port — or attach an Auto Scale Group — and set the health check and routing policy.
Secure the frontend
Attach an imported SSL certificate, enable HTTP→HTTPS redirect, add security groups, and optionally LUKS encryption at rest and BitNinja protection.
Deploy & observe
Review the summary, deploy, and point DNS at the LB IP (reservable). Watch live stats, alerts and logs from the same console.
Wired into the rest of the platform
A load balancer is rarely alone — E2E's connects natively to the services around it.
Compute & GPU nodes
Balance across CPU compute nodes or GPU nodes serving inference — mixed backend groups, WebSocket timeouts included.
VPC, IPv6 & reserved IPs
Attach up to three VPCs, run fully internal LBs for private tiers, reserve your public IP, and attach IPv6 when you need it.
Kubernetes
Running containers instead? E2E Managed Kubernetes ships with MetalLB-based LoadBalancer services and reservable service IPs out of the box.
Frequently Asked Questions
Everything you need to know about Load Balancer.
Load Balancer
Each load balancer is a dedicated managed appliance on a plan you choose (vCPU, RAM, storage, max concurrent connections), with its own IP, security groups and optional LUKS encryption at rest — managed and monitored by E2E.
Import your certificate with its private key and chain into the SSL Certificate Manager, then select it when the frontend protocol is HTTPS or Both. You can force HTTP→HTTPS redirect and use the HTTPS backend protocol to keep traffic encrypted all the way to your servers. A certificate attached to an LB can't be deleted until detached.
ALB health checks mark it DOWN and stop routing to it until it returns 2xx/3xx again. You can watch per-server UP/DOWN status in Backend Mapping and live session/error counters in the Stats tab, and set alerts on resource thresholds.
Yes — convert to a Committed Load Balancer for committed-term billing with auto-renewal options. Hourly billing remains the default for everything else. Prices on the rate card.
Yes — ALB backend groups include a configurable WebSocket timeout, alongside tunable connection, client, server and HTTP keep-alive timeouts.
Put a managed LB in front of it
Deploy in minutes — no sales call required.