Internal Load Balancing in 5 mins
Deliver scalable and resilient internal-only services on GCP
Google Cloud Load Balancing
HTTP(S) Load
Global SSL proxy
Balancing
Internal TCP/UDP Network TCP/UDP
Regional
Load Balancing Load Balancing
Hash-based LB Capacity-based LB Global Load
Request Routing
algorithm algorithm Balancing
Connection Cross-region
Health Checks Autoscaling
Draining failover
Affinity Logging Monitoring Cloud CDN
Internal Load Balancing
LB IP:
10.10.10.1 Src IP: 10.10.10.1 Src IP: 10.10.10.1
Client A Backend B
Dst IP: 10.10.10.200 Dst IP: 10.10.10.200
X
Internal (RFC 1918) Client IP preserved Health checks Autoscaling Session Affinity
Load Balancing (TCP, HTTP, HTTPS) (no prewarming)
REST API gcloud CLI Console
Internal Load Balancing Example
Subnet 1 Subnet 2 Subnet 3
Internal LB IP
Client instances 10.240.0.200
10.240.0.1 10.240.0.2
192.168.1.5 Zone: us-central1-b
192.168.1.1
Zone: us-central1-a
172.16.1.1
10.240.0.3
Zone: us-central1-c
Network: shopnet
Google Cloud Platform US Central region
Logical representation of the ILB, there is no LB instance between client and backend instances Backends running shopping cart application
Internal Load Balancing
Under the hood: Software-defined load balancing is delivered using
Andromeda, GCP’s
software-defined
Load Balancers can become Andromeda network virtualization
choke points stack
There is no load balancer in
the path between your client
and backend instances
Client Client Client Client
instance ...... instance instance ...... instance
IP1 IP2 IP1 IP2
Virtual Network
VIP
Load Balancer
instance(s) Internal LB
IP3 IP4 IP5 VIP VIP VIP
Backend Backend Backend Backend Backend
Backend ...... instance instance ...... instance
IP3 IP4 IP5
Typical Instance-based Internal Load Balancing Software-defined Internal Load Balancing on GCP
Configuring Internal Load Balancing
(console)
Infrastructure & Operations 6
Internal Load Balancing Configuration
Client
Instance
Forwarding Rule Internal LB IP, port
configuration
Regional Backend Service
Backend 1 Backend 2 Backend 3
Backend Configuration
Instance Group 1
Instance Group 2
Instance Group 3
Instance 1 Instance 3 Instance 5
Instance 2 Instance 4 Instance 6
Subnet A Subnet B
Region X
1. Click “create load balancer” to get started
Click Create
2. Select the type of Internal Load Balancer- either TCP or UDP
Click here for TCP
ILB
3. Specify you want to load balance traffic between your instances (internal)
Specify you want internal
load balancing
4. Configure backends in your region of choice
Configure backends
Remember to open up firewall
for ILB health checks
5. Configure your RFC 1918 Internal LB IP (specify or let ILB auto-allocate)
Configure Internal Load
Balancing IP and port (or
list of ports)
6. Click “create” and your ILB is ready to distribute traffic!
Click Create.
Done!
Internal Load Balancing Use Cases
Infrastructure & Operations 14
ILB Use Case 1: Scaling and HA for internal (micro)services
Load balance across multiple
instances of your internal service
Internal Load Balancer Internal Load Balancer
Internal Load Balancer
RECOMMEND RECOMMEND
SHOPPPING SHOPPPING SHOPPING
CART CART CART
CATALOG CATALOG
Google Cloud Platform Application composed of ILB-scaled services
ILB Use Case 2: Multi-tier apps Internal Load Balancing us-central-1a
Web Tier with external load
balancing Database Tier
IP: 10.10.10.10, Port: 80
Internal Load Balancing
Internal Tier
Web Tier
External Load Balancing
Global: HTTP(S) LB, SSL Proxy
User in San Francisco Regional: Network TCP/UDP LB
Database Tier
Web Tier Internal Tier
User in Iowa 100.1.1.1, Port: 80 us-central-1b
Internal Load Balancing
IP: 10.20.1.1,
User in Singapore
Port: 80
Internal Tier with Internal Load
Balancing Database Tier
Web Tier
Internal Tier
asia-east-1a
ILB Use Case 3: Scale-out and HA for virtual appliances
Internal Load Balancer
10.0.0.2 10.0.0.3 10.0.0.4
Virtual Appliance Virtual Appliance Virtual Appliance
Instance 1 Instance 2 Instance 3
Google Cloud Platform
Learn more
https://cloud.google.com/compute/docs/load-balancing/internal/
‹#›
