HA - High Availability

When it comes to availability, really what we're talking about is high availability: making sure that your systems are up and running for your users. You're going to have some downtimes in systems, it's really rare for us to not have any downtime in our systems, but trying to maintain a high level of availability for our users and limit how much downtime we have is a big key factor in how we're going to be successful with the CIA triad.

Availability is obviously a key component, otherwise it wouldn't be part of the CIA triad. That is, we are going to frustrate our customers and our users, or whoever is using our systems, if we don't maintain a certain amount of availability.

The impact of downtime

When we have a lack of availability, that is, there is downtime on our systems, it can really have a wide range of impact to the business. But quite often what I find is that there's a loss of revenue if you're an e-commerce site, there's a loss of reputation that happens, and eventually, if it happens often enough, you have a loss of customers. So although it can affect other areas, those are the three areas I find it most impactful.

The impact of downtime or lack of availability on your system really depends on a few different things, and on how the customer or users are viewing the situation, and that could be a negative impact. For instance, how often it happens. This is a timeline right here, and let's say you have an outage right there, a downtime, a time when your systems are not available. That can have an impact. And then let's say it happens again, and then again, and then again. What happens is that you have a loss of reputation over time, and people are going to grow weary of that, are going to get tired of this scenario of the systems always being down, and they're going to give up on the company or give up on the services. So that's one impact. The other thing that can happen is if there's extended impact, if these get drawn out for too long, or even just a single one that's drawn out for a long period of time, that can be really problematic as well.

Availability isn't necessarily an all or nothing thing either, though. That is, we have performance and partial down. For instance, you could have performance issues. Maybe you haven't gone completely down, maybe your services are still available, however they're not really available to the level of performance that you really need, and so there's a performance issue. That still is going to frustrate your users, and how often that happens and how long those durations are is going to be impactful to the customer. You also have a partial down, where maybe not all of your services are down, maybe there's just select services that are down, that are going to cause problems.

Threats to availability

So why would you have a lack of availability? One thing that we've mentioned already is that there are threat agents out there. There are people actively trying to bring your systems down; that's called a denial of service attack. Perhaps they're trying to sabotage you or take you out of business. Maybe it's a competitor that's trying to bring down the level of competitiveness that you have. Maybe it's some sort of theft of equipment that really was needed to keep your systems up and running.

But there's other things as well. There's utility disruptions, where the internet goes down or power goes down. There are failures of hardware, software errors that come up. There is destruction that can happen when fire or earthquake or some sort of natural disaster takes place. And then I find more often than not availability really goes down a lot of times because of human error, a misconfiguration that brings the system down.

What high availability is

So when it comes to availability, really what we're looking for is high availability. What is high availability? Let's talk about it from a user's perspective. A user gets access to your system. Let's say you host some sort of application through the web to your users, so your users get online onto your system through the internet, and then you have a back end database to this right here. Really what high availability is, is that it's more subjective: it's really how your users perceive how available your systems and services are.

Faults, downtime and fault tolerance

So one thing that can happen on your network is a fault, and a fault could have the potential to limit how much availability you have to your users. Let's give an example of this. Let's say there is software on this web server right here that's hosting out the web services, and there's an error on that, and so that service stops working. That's a fault. What that does is it creates a downtime. A downtime is the time that your users don't have access to your services.

So what we want to do is create a system that is fault tolerant. That means that if there's a fault on it, where a service goes offline, there's other elements that are in place here that will take over. That is fault tolerance: that this system is tolerant to any faults, that they can compensate for any faults on the system.

One big way we do that is with redundancy. So instead of just having one server running our web application, we have a bunch of servers that are running our web application, and so if one goes offline another one can pick it up. Same thing with the database: let's say we don't have just one database server up and running, maybe we have multiple database services up and running, and so if one of these goes offline the other one steps in and takes over.

Limiting downtime

So really, how do we limit downtime? What's the big picture here? One is we need to eliminate weak points. An example of a weak point is a single server, because things go offline. So how are we going to compensate for that? One way is to create redundancy on our network.

We can also implement things like monitoring and alerting. Things that monitor our systems on whether they're up or down and what's happening on our networks does two things. Number one, it can show us early signs that something is going off, that something is wrong on the network, and hopefully we can fix it before it becomes an actual downtime, before it becomes an actual problem for our customers. The other thing is that if there is a problem, if there is something that goes wrong on our network, it alerts us quickly so we can step in and start fixing it. And that's the response side of this: how are we going to respond, and making sure that we respond in a quick way.

And then finally, also the planning side of this, which helps us get good at the rest of this, is that we actually do planning for if there is downtime, if there is an issue, how are we going to respond to it.

Plans you can put in place

So one way that we can limit the amount of downtime that we have and increase the amount of availability is by planning ahead of time and having certain plans in place. There are several different plans that we could put in place that could help us out.

One of them is an incident response plan. An incident is when something happens: a server goes down, the internet connection goes down, maybe our database goes down, something goes down and we need to respond to that. So what we're going to do is have a set of who's going to be in charge, how are they going to approach troubleshooting, how are we going to do communication, who do we need to notify. That's all part of an incident response plan.

A disaster recovery plan takes this to an even bigger level of, what if our whole data center goes down? There's an earthquake and all of our data center gets wiped out. What is our plan at that point in time?

A business continuity plan takes it from an even broader perspective: how do we continue operating as a business when things go offline? Where a disaster recovery plan says, if there's an earthquake and our data center goes down, how do we recover all of our data and get back up and running, a business continuity plan says, what happens when we have half of our employees that are offline because their connection is down and they can't get into the office, and how are we going to overcome this earthquake that's happened from a more global perspective?

One other term that we could hear is something called a contingency plan, or some will call it a continuity of operations if it's not actually a business. So if it's like a federal government, they'll call it a continuity of operations because it's not a business.