A cloud administrator observes a holiday spike in traffic to a site hosted on an IaaS platform with a load balancer and standardized VMs. Which option would be most efficient, with no downtime, to handle the surge?

A cloud administrator observes a holiday spike in traffic to a site hosted on an IaaS platform with a load balancer and standardized VMs. Which option would be most efficient, with no downtime, to handle the surge?

• A. Implement an upward vertical scaling solution.
• B. Implement a downward vertical scaling solution.
• C. Implement an inward horizontal scaling solution.
• D. Implement an outward horizontal scaling solution.

Answer: (A)

When a sudden traffic spike hits an IaaS deployment behind a load balancer, the fastest way to absorb the extra load without taking the site offline is to increase the resources of the existing virtual machines (scale up). By boosting CPU and memory on the current VMs, you immediately provide more capacity per host, so each request can be processed faster without introducing new servers or changing the network configuration. If the cloud platform supports live resizing, this can be done with no downtime at all; if not, any brief downtime is typically shorter than what would be involved in provisioning and bringing up new instances. Scaling out by adding more VMs behind the load balancer is a solid strategy for sustained growth and avoids hitting per-VM limits, but it requires provisioning new instances, waiting for them to come online, and integrating them into the load balancer pool, which can introduce delays. Scaling downward reduces capacity and isn’t helpful for handling a surge. Scaling in or scaling out are both valid strategies in different contexts, but for the fastest, zero-downtime response in this scenario, increasing the resources of the existing VMs is the best fit.

When a sudden traffic spike hits an IaaS deployment behind a load balancer, the fastest way to absorb the extra load without taking the site offline is to increase the resources of the existing virtual machines (scale up). By boosting CPU and memory on the current VMs, you immediately provide more capacity per host, so each request can be processed faster without introducing new servers or changing the network configuration. If the cloud platform supports live resizing, this can be done with no downtime at all; if not, any brief downtime is typically shorter than what would be involved in provisioning and bringing up new instances.

Scaling out by adding more VMs behind the load balancer is a solid strategy for sustained growth and avoids hitting per-VM limits, but it requires provisioning new instances, waiting for them to come online, and integrating them into the load balancer pool, which can introduce delays. Scaling downward reduces capacity and isn’t helpful for handling a surge. Scaling in or scaling out are both valid strategies in different contexts, but for the fastest, zero-downtime response in this scenario, increasing the resources of the existing VMs is the best fit.