CISSP 9.5 - Server-Based Systems

This episode of the ISC2 Certified Information Systems Security Professional (CISSP) exam prep series looks at how servers move data and share heavy workloads, continuing Domain 3. A server that cannot manage its own traffic will collapse under load, whether from an attack or its own success — this session covers how that flood is throttled, spread, and absorbed across different computing models.

Frequently Asked Questions

What is data flow control, and why does it matter?

Data flow control is the management of how data moves between processes, across networks, and over channels. Done well, it keeps transmission efficient with low latency, reliable through integrity checks, and confidential through encryption, and it keeps receiving systems from being swamped past the point where they drop connections. When data overflows, it can be lost, corrupted, or forced into costly retransmission, and flow control is provided by devices like routers and switches as well as network services.

How does a load balancer help?

A load balancer spreads incoming traffic across several links or servers so no single one is overwhelmed, delivering better use of your infrastructure, faster response, higher throughput, and no bottlenecks. A common deployment fans requests out across the members of a server farm, using techniques like round robin, random choice, or watching real-time utilization to decide where each request goes.

What happens when traffic is malicious?

You face a denial of service attack, which can cripple data flow control outright. By flooding a system with bogus traffic, an attacker aims to exhaust its capacity so legitimate users cannot get through. This is why you monitor for these attacks and build in mitigations before they strike.

How do parallel systems split enormous work?

They break a huge task into smaller pieces and hand each piece to a different processor to compute at the same time. In symmetric multiprocessing, several processors share one operating system, one data bus, and common memory, working together on a single main task; in asymmetric multiprocessing, each processor often runs its own operating system and owns its own memory, and specific tasks can be pinned to specific processors, an arrangement called affinity. A more extreme variation links many asymmetric systems into a massive cluster with hundreds or thousands of processors for the heaviest scientific and mathematical workloads.

What makes grid computing risky?

Grid computing loosely pools many volunteer nodes that join and leave freely, and that openness is exactly the problem. Because anyone can participate, the contents of each work packet may be exposed to the world and members could keep and inspect copies, so grids are a poor fit for private or proprietary data. Capacity also swings wildly, packets can come back late, corrupted, or not at all, and because a grid leans on central servers to coordinate, knocking those out can bring the whole thing down.

Reference: This article is based on concepts discussed in CISSP 9.5 - Server-Based Systems.