We use analytics cookies.We use cookies to analyze site traffic and improve your experience.
Loading...
System Design and AI Engineering, by Roni Das. 984 interactive lessons.
Start LearningFoundation · Load Balancing & Proxies
A load balancer that routes users to the nearest data center or region based on geography, latency, or health. DNS-based (Route 53, Cloudflare) or anycast-based.
In short
A global load balancer is a traffic director that spreads user requests across data centers in different regions, sending each user to the closest or healthiest one. It works at the global level using DNS routing (like AWS Route 53) or anycast IP routing (like Cloudflare and Google Cloud), unlike a regional load balancer that only spreads traffic across servers inside a single data center.
In short
A global load balancer is a traffic director that spreads user requests across data centers in different regions, sending each user to the closest or healthiest one. It works at the global level using DNS routing (like AWS Route 53) or anycast IP routing (like Cloudflare and Google Cloud), unlike a regional load balancer that only spreads traffic across servers inside a single data center.
Learn it in the course
This page is the free part.
₹1,111 in India$25 everywhere elseonce, for the whole course
The System Design course covers Global Load Balancer across a run of lessons, not one page. These 4 alone are about 134 minutes of step-by-step reading, every one with a quiz.
How a GSLB decides which region a user reaches, and why the decision takes minutes to arrive even when it is made in microseconds.
Latency-based routing sends each user to the region with the lowest measured round trip. Route 53 latency records, geolocation vs geoproximity, and a real lab.
How load balancers decide which server gets the next request. Round Robin, Weighted, Least Connections, IP Hash, and more.
How load balancer failover works with a shared virtual IP, VRRP and keepalived. Also covers ALB health checks and failover across zones.
all part of
System Design Masterclass
770 lessons · about 287 hours · 18 free to read
₹1,111in India, by UPI
$25everywhere else, by PayPal
It is really nice and good Please join the course
You own this course
Continue with Global Server Load Balancing (GSLB)pay once,
yours for life
A normal load balancer sits in front of a pool of servers in one data center and decides which server handles each request. A global load balancer works one level higher: it decides which data center (or cloud region) a user reaches in the first place. If you run servers in Virginia, Frankfurt, and Singapore, the global load balancer is what sends a user in Berlin to Frankfurt instead of bouncing their request across the Atlantic.
The two main jobs are reducing latency and surviving failures. Latency drops because the user talks to a nearby region instead of a distant one. Failover happens because if the Frankfurt region goes down, the global load balancer stops sending traffic there and routes those users to the next best healthy region without anyone changing their bookmarks.
It also handles uneven load. If one region is overloaded during a regional traffic spike, the balancer can shift a portion of requests to a less busy region that still meets a latency target.
There are two common mechanisms. The first is DNS based. When a user's browser looks up your domain, the DNS service returns a different IP address depending on where the user is and which regions are healthy. AWS Route 53 and older GSLB appliances work this way. It is simple and works with any protocol, but DNS answers get cached by resolvers and devices, so a failover can take seconds to minutes to fully take effect because old cached IPs keep pointing users at a dead region until the TTL expires.
The second is anycast based. You announce the same IP address from many locations using BGP, the internet's routing protocol. The network itself delivers each user's packets to the nearest location announcing that IP. Cloudflare, Google Cloud's global external load balancer, and AWS Global Accelerator use anycast. Failover is faster because rerouting happens at the network layer in seconds, and there is no DNS cache to wait on.
Both approaches lean on continuous health checks. The balancer probes each region's endpoint every few seconds, and a region that fails its checks is pulled out of rotation. Routing decisions can be based on geography, measured latency, weighted percentages, or capacity headroom.
Reach for a global load balancer once you serve users across continents and run application servers in more than one region. If everything lives in one region, a regional load balancer is enough and a global one just adds cost and moving parts.
The big trade-off is consistency. Sending users to whichever region is closest is easy for stateless reads. It gets hard the moment users write data, because now you need cross-region replication and you have to decide what happens when a user in Frankfurt and a user in Singapore edit the same record. Global load balancing solves traffic routing, not data consistency, and people often confuse the two.
DNS based balancing is cheap and protocol agnostic but reacts slowly to outages. Anycast reacts fast but needs control of IP ranges and BGP, which is why most teams rent it from a provider rather than running it themselves. Either way you pay for redundant capacity in multiple regions that mostly sits partly idle.
Picture a streaming app with regions in us-east-1, eu-central-1, and ap-southeast-1, fronted by Cloudflare anycast. A user in London opens the app. Their packets hit Cloudflare's London edge, which forwards them to eu-central-1 in Frankfurt, the nearest healthy origin region. Round trip latency stays around 20 to 40 ms instead of the roughly 80 ms it would cost to reach Virginia.
Now eu-central-1 has an outage. Health checks fail within a few seconds, the global load balancer drops Frankfurt from rotation, and the London user's next requests get routed to the next closest healthy region, say us-east-1. Latency rises to roughly 80 ms but the app stays up, and the user never sees an error page or has to do anything.
Cloudflare
Runs anycast across 330+ cities so the same IP reaches the nearest edge, with automatic failover to healthy origins.
AWS Route 53
DNS based global routing with latency, geolocation, and weighted policies plus health checks to fail traffic away from down regions.
Google Cloud
Its global external load balancer uses a single anycast IP backed by Google's network to route users to the closest healthy backend region.
AWS Global Accelerator
Provides static anycast IPs that move user traffic onto the AWS backbone at the nearest edge and reroute on regional failure in seconds.
A regional load balancer spreads requests across servers inside one data center or region. A global load balancer sits above that and decides which region a user reaches in the first place, based on geography, latency, or which regions are healthy. Most real systems use both together.
No. A CDN caches and serves static content (images, scripts, video) from edge locations close to users. A global load balancer routes live application traffic to your origin regions. They are often used together, and providers like Cloudflare offer both, but a CDN serving cached files is doing a different job than routing dynamic requests to the right backend.
DNS based (such as Route 53) is simple, works with any protocol, and is cheap, but failover is slow because resolvers cache the DNS answer until its TTL expires. Anycast based (Cloudflare, Google Cloud, AWS Global Accelerator) fails over in seconds because rerouting happens in the network layer, but it requires control of IP ranges and BGP, so most teams rent it from a provider.
No, and this is a common mistake. It only routes traffic to a region. Once users in different regions write data, you still need cross region replication and a strategy for handling conflicting writes. Routing is solved separately from keeping data consistent everywhere.
With anycast, network rerouting plus health checks usually take effect within seconds. With DNS based routing it can take from tens of seconds to several minutes, because cached DNS records keep pointing users at the failed region until the TTL expires, even after the balancer marks it unhealthy.
DNS
The phonebook of the internet. Translates human-readable domain names (google.com) into IP addresses that computers understand.
CDN
A network of servers distributed globally that caches content close to users. Netflix uses CDNs to stream video from servers near you, not from one central location.
Cloud Region
A geographic area containing one or more data centers (availability zones). Choosing the right region reduces latency and satisfies data residency requirements.
Proxy
An intermediary server that sits between the client and the destination server. Forward proxies act on behalf of clients; reverse proxies act on behalf of servers.
Reverse Proxy
A server in front of your web servers that receives every request first, passes it to one of them and returns the answer. It handles TLS, load balancing, caching and routing, and hides the servers behind it.
DNS Record Types
Different types of DNS records map domains to resources: A records point to IPs, CNAME aliases one domain to another, MX routes email, TXT stores verification data.
This page explains the idea. The "Global Server Load Balancing (GSLB)" lesson builds it up step by step, with diagrams. It is part of the System Design Masterclass course, which has 770 lessons. 18 of them are free to read. You pay once and keep it for life, with no subscription.
Open the Global Server Load Balancing (GSLB) lessoncourse 1
From absolute beginner to principal engineer, drawn step by step.
Asked about Global Load Balancer in an interview?
Each question has why it is asked, a full answer, what the interviewer listens for, the follow-ups, the common mistakes and a diagram (268 diagrams in all). A PDF you keep, junior to staff level.
India: UPI, cards or netbanking through Razorpay. Elsewhere: card or PayPal. Download at once from My packs.
