Keeping data secure on our individual systems is absolutely critical. And it's probably the cornerstone of all IT security there is. But let's make sure we understand what we're talking about when we talk about securing data. Now, in this case, what I'm avoiding is permissions based issues. I'm not talking about how do we secure data to keep somebody from editing a file or something like that. What I'm talking about is just within the system itself, how do I keep all the data happy.
So what we're really talking about here is number one, data integrity of some form or another, making sure that if a hard drive dies, or if a power supply dies or something like that, I still have that data. We're also talking about speed. So we want quick access. So generically, when we're talking about these types of things. What we're talking about is what I'm going to call high availability. So we'll save a keeping out the hackers and keeping out the malware for other episodes.
What we're talking about is the pure system, what can we do to keep that data happy. So let's think about high availability. So probably the best place to start is raid. raid is a great way to secure data. Depending on the RAID level you use, you can provide good integrity, you can provide good speed. And the nice part is, is raid nine times out of 10 is one of the cheapest ways you can go.
Raid is a fantastic tool. But raid is only good for the drives or the storage within a system. So if you lose a hard drive raid is a great way to take care of these types of issues. However, what happens if I lose a power supply? What happens if the entire server explodes? Now I've got a really big problem.
So when we're talking about securing data, we need to stop thinking about just the drive storage and start thinking about the systems as a whole, then probably the best most core way we do that is something called clustering. clustering is a simple concept. So here I have a single computer that might have raid in it might not don't know clustering simply means instead of having one computer doing the job, you have two computers or more doing the exact same job sharing the same database sharing whatever it might be. The nice part about clustering is that if one of these systems dies, the other one can automatically kick in. The downside to clustering is that they constantly have to keep each other updated. clustering is absolutely fantastic.
In fact, you can see some pretty interesting setups where two servers that are acting as clustering servers will literally have their own dedicated little network between them because they have to update each other so much. However, note number one, clustering is really expensive to literally not just recreate a couple of extra drives, but to recreate an entire system can add to the price of securing data dramatically. The downside to clustering also is that you really usually have one primary cluster machine and that way, he's always doing all the work. You'll have other machines that act more like a backup system. So we could really use these systems a little bit more efficiently. And when you do that you get load balancing.
Here are my two clustered servers. Now normally with clustering, one is the primary and one is kind of like the backup. However, with load balancing, all you're doing is you're distributing the workload across all of these clustered servers. That way, no one machine is totally overwhelmed. clustering and load balancing are absolutely powerful tools, but they do tend to be a little pricey. So one of the things we also like to do for securing data is to simply virtualize the servers.
In that case, if a server were to die in one fashion or another, we can simply bring it back from a snapshot. Now, just because the servers virtualized don't think that you don't still have the ability to do raid with virtualized drives, or to do clustering with virtualized servers. But really when it comes to the ultimate in securing your data virtualizing this stuff really pays off.