The NSX Advanced Load Balancer Global Server Load Balancing (GSLB) balances the application load across instances of the same application deployed to multiple locations in data centers or public clouds.
GSLB Overview
The GSLB directs the client requests to the closest or best-performing data center (as per the configured GSLB algorithm) or the surviving data centers (if outage occurs). Application load at any of those locations is usually managed by a local load balancer, which might be the NSX Advanced Load Balancer or a third-party ADC (Application Delivery Controller) solution.
GSLB is implemented to achieve the following application goals.
Provide optimal application experience to users or clients from geographically distributed areas.
Offer resilience when loss of a data center or a network connection occurs.
Perform non-disruptive migration to another data center or addition of a data center.
NSX Advanced Load Balancer GSLB Overview
GSLB includes the following functions.
It selects the location, namely a data center or cloud, to which the requests of the client will be directed.
It monitors the health of the application instances to select the best location and rule out unhealthy ones.
It synchronizes configuration and state across GSLB sites so that functions 1 and 2 can continue despite certain failures.
When a client, typically a browser, performs a DNS query on Fully Qualified Domain Names (FQDNs), GSLB responds with the IP address (VIP) of the optimal application instance. The optimal address can change based on the load balancing algorithm, health of the application instances, and the location of the clients.
Use Cases for GSLB
The following are a few use cases for NSX Advanced Load Balancer GSLB.
Optimal application experience for geographically distributed users (Active-Active GSLB configuration)
Applications are deployed in multiple data centers.
GSLB can redirect user request to the most optimal location. This helps enhance the user experience, deliver the localized content, and so on.
Application high availability across data centers (Active-Active GSLB configuration)
Applications are deployed in multiple data centers.
If a data center failure occurs, application instances running in the remaining data centers can take over the user traffic.
It leads to increased reliability.
Disaster recovery (Active-Passive or StandBy GSLB configuration)
Applications are deployed in two data centers.
While both are healthy, all traffic is directed to the primary DC.
If the primary DC fails, the global DNS directs all user traffic to the other. This is done with priority algorithm in NSX Advanced Load Balancer GSLB.
Hybrid cloud with cloud bursting
Applications are deployed across on-prem, private, and public clouds.
On-prem mode is preferred and configured with a higher priority. If the on-prem pool goes down, traffic can be sent to the configured public cloud as pool of lower priority.
In addition, min-members configuration knob can provide a way in which pools of different priorities can be used, if some of the on-prem members are down. For more information, see Minimum number of Servers.
Both on-prem and cloud members can be added to a single pool and weights can be added to prioritize one over the other.
How NSX Advanced Load Balancer GSLB Works
To understand how GSLB works, consider an environment as illustrated in the following figure.
In this environment, the NSX Advanced Load Balancer runs in four locations (GSLB sites), three of which are on-premises, and one in Amazon Web Services (AWS). Each site has its own NSX Advanced Load Balancer Controller Cluster (represented by a single Controller icon).
Application A has virtual services running in all four locations. These virtual services are identified by VS-A1 through VS-A4.
Three of the four locations (DC-1, DC-2, and AWS) have global DNS services (DNS-1, DNS-2, and DNS4) that are synchronized. They are all equally authoritative for the subdomain A.acme.com.
The fourth site (DC-3) does not run a global DNS service.
The above image is for illustration purposes only. The image depicts DNS running at specific locations. Based on the requirement, you can decide if you want to run the DNS services on one site, multiple sites or all the sites.
