Reliability? Diversity....

Mushy-pea 0 Tallied Votes 268 Views Share

I was in a pub in Manchester recently for my friend Chris's leaving do (he's got a job in Spain :p ). Somehow a brief discussion about Mac OS X got started (yes, OK I started it). Most people I've met who've used it seem to think it's great because:

1. The GUI is great.
2. It's really reliable.

This issue of reliability and security is a big one in the IT scene. Most users, especially those managing large corporate IT systems, would like an operating system to be virus and hacker proof. Being uncrashable would be a bonus as well.

Thinking about these issues reminded me of the Airbus A320. As you may know, it is too aerodynamically unstable to be flown directly by humans. The pilot operates the controls and a computer interprets these "suggestions" about how the plane should be flying. It does some very complicated maths and works out how to fly the plane safely. However, there are actually four computers doing this job simultaneously. Each one has a different hardware architechture made by a different manufacturer, with a different operating system from a different developer running each machine. The same goes for the flight control software. A failure in one machine should instantly trigger the switching of control to the next one.

This design was chosen to minimise the horrific chance that a computer crash would cause a....CRASH. It seems that many companies choose one OS for a particular application and put their trust in that. I know the company hosting my website uses only Linux servers. So, what happens if somone finally manages to develop a really nasty and effective virus for Linux? Goodbye to my website (for a while anyway). Also goodbye to a few million other sites. I don't have anything against Linux. The same could be said about companies using only Windows on their desktop machines.

I don't know what the ideal solution to reliability is. Obviously using multiple redundant mirror servers with different OS's is costly and possibly tricky to implement. But you get the picture....the diverse will survive. In my opinion.

Steven.

Dani AI

Generated

Both posts raise the real trade-off: a single-platform environment is cheap and easy to run, while diversity reduces the risk that one exploit topples everything. highlights the vulnerability of monocultures, and points toward keeping an alternate recovery option — both are sensible starting points.

Practical guidance is to treat diversity as one tool in a wider resilience strategy, not the whole answer. Use layered defenses (network segmentation, least privilege, application hardening), automate image building and patching so multiple stacks stay consistent, and push redundancy only where the impact justifies the added operational cost. Virtualization and immutable images speed recovery; orchestration and automation make multi-stack setups manageable.

Immediate, actionable checklist:

  • Inventory and classify assets by business impact.
  • Find single points of failure and decide where true stack diversity is worth the cost.
  • Standardize image creation and provisioning so builds are reproducible across OSes or hypervisors.
  • Keep tested backups, golden images, and automated restore procedures; run failover drills.
  • Segment networks and apply host-level controls to limit lateral movement.
  • Stagger patch rollouts and use canary hosts to catch regressions before full deployment.
  • Ensure staff skills and runbooks match the number of supported stacks.

Maintaining an alternate boot or fallback image, as suggested in essence, is useful — but treat it as part of a practiced recovery plan. Without regular testing and monitoring, any fallback can fail when you need it most.

John A 1,896 Vampirical Lurker Team Colleague

You do have very good points about viruses, however, a lot of companies not only use only one operating system because they believe it's more stable, but it's much cheaper to buy 1 copy and network license for all their computers (commerical OS), and it's also much cheaper to maintain computers that all have the same operating system.

A cheaper solution for this might be to create a dual-boot system with multiple operating systems, and keep the data between the operating systems mirrored. Then if a problem arises, the company can resort to their backup operating system(s). This is assuming, of course, that there's hard disk space to spare.

Be a part of the DaniWeb community

We're a friendly, industry-focused community of developers, IT pros, digital marketers, and technology enthusiasts meeting, networking, learning, and sharing knowledge.