Redundancy is not a new concept, but not all redundancy is equal. The value of redundancy depends on how closely (or broadly) you look at it.
If you have five storage devices working in RAID 5 you may have one of those storage devices fail, but all of your data would remain intact due to the support of the redundant drives.
To use an analogy, if you have five warehouses and they are all are filled to 80% capacity, you can move some of the contents into the extra space left in all the other warehouses in the event that one experiences a flood. Normally, this means each warehouse uses less than 100% of the available capacity, so in the event of a flood, you can redistribute the contents of the flooded warehouse into the undamaged warehouses. This redistribution allows you to keep working on cleaning up the debris from the damaged warehouse while keeping your goods intact.
Redundant devices work just like the extra warehouses. When one device fails and must be taken out for repair, the other devices keep your data safe. But what happens if you only have one drive controller?
Think of it as if your inventory system, responsible for organizing all those warehouses, gets destroyed. The warehouse still has all of its items located inside, but you have no idea where things are within the building. To prevent this, there should always be at least two units for all significant hardware devices that enable smaller components to function.
At Purely Cloud, our first concern is always the client experience. That’s why all our significant components have at least 2N redundancy. We’re even currently making the move to 2N+1 redundancy for vital components.
However, this is only the direct hardware side of the redundancy equation.
Hardware and Software Interactions
Let’s talk a little bit about hardware and software interactions.
On Purely Cloud, there are many servers working together to execute an incredibly wide variety of tasks. Each of these servers can be thought of like an employee that has tasks that need to be completed.
Many companies will load their infrastructure in order to maximize profit; meaning they have only just enough hardware to execute the current load (N). Additionally, if any single server has an outage, all tenants will feel the impact.
Keeping with our analogy, if a single employee has a sick day, it negatively impacts all of the organization’s clients. Most people staff their company to accommodate the realities of life. Similarly, a large number of providers set up their infrastructure to have some tolerance for minor failures. This is done by having a single server on standby or by not loading servers over the amount a single server outage could accommodate.
This practice is like having one employee that is specifically hired to cover for the absences of other employees when they are unavailable. However, what would happen if there was an illness going around and half your team had to go to the hospital for a few days?
This is, essentially, what could happen if you lost one of those two vital components we talked about earlier. Very few organizations are prepared to encounter such an event.
Purely Cloud Can Help You Prepare for the Worst
At Purely Cloud we don’t think our bottom line is defined by how quickly we can increase our profit margins, but by how many clients we can give a revolutionary experience.
We load our infrastructure to a level of 2N+1, which means: we can have half our cloud “team” be in the hospital while someone on the remaining team had a sprained ankle, and you would never even know the difference.
We think the less you talk to us about the cloud and the more you want to talk to us because we’re people you love to interact with, the better.
Learn More About the Cloud at nVerge In the Cloud
Purely Cloud is a proud Diamond Sponsor of nVerge In the Cloud, FMT’s free online technology conference. This year at nVerge there will be 14 great sessions that you won’t want to miss.
Check out the full agenda, and register for sessions, here.
Business Development Manager, Purely Cloud