Join Ed Liberman for an in-depth discussion in this video Deploying primary storage devices, part of CompTIA Server+ (SK0-004) Cert Prep: 3 Storage.
- [Instructor] When anticipating the installation of primary storage devices on your servers, there are several factors that you should consider before you even go out and purchase these devices. The first of these factors is understanding that there are a number of disk specifications that you need to be familiar with. Now I want to tell you that if you're out shopping for these storage devices, the number of technical specifications that are listed can be quite overwhelming.
But if you take the time to understand each one individually, it's really not that difficult to understand and know how to purchase the correct device. So the first of these specifications is RPM, which stands for revolutions per minute. Now I will tell you that this is specific to an actual hard drive, not an SSD, which is a solid state drive, which doesn't have any moving parts, okay? An actual traditional hard drive where you have disks that are spinning, well RPMs is how many revolutions those disks spin per minute.
I want you to keep in the back of your mind that RPMs have an inverse relationship with another specification that we're going to talk about in just a few minutes here, which is called latency. What I mean by inverse relationship is the larger number of RPMs, the smaller the latency number is going to be, and I will explain when we get to latency why that is. Now next we have the dimensions or the form factor of the hard drive. Now this might seem like the simplest, and the one to not pay any attention to.
There's not a whole lot of variables here. I mean typically you have a 2.5 inch small form factor or a 3.5 inch large form factor. Those are the two primary ones to pay attention to, but here's why it's important to pay attention. If you don't any you only look at all the other specifications, and you buy what theoretically is the perfect device, only to get it back and realize it doesn't fit into your server, well then you still have a problem. So you want to make sure that you know the bays that are available, whether they are 2.5 inch or 3.5 inch when you're purchasing the drive.
Capacity, this is an extremely important, and I would say for most people the obviously important specification. Not a very difficult one to understand. It is how much data the drive can hold. And, when you are purchasing drives, well you want to make sure that you purchase enough storage capacity for your needs. The one thing that I will mention is that when you are looking at capacities you want to also look at the prices and look for what we call the sweet spot.
Now this sweet spot is a moving target, okay? It's always changing, based upon the technologies that are out at the moment that you are purchasing. But you will find that as you get bigger drives, they will become more expensive, but as you go larger and larger, it will only become a little bit more expensive, but then you'll hit some point where there'll be a big jump, and very often what you can find is that it's more cost effective to purchase multiple smaller drives that equal the storage capacities that you need rather than purchase one large drive, which costs way too much money.
Now the next specification we have here is what's called the bus width, and I'm going to tell you that for the most part this is a specification that you don't need to worry about much anymore. The reason why is because most of today's devices are more commonly serial attached SCSI or serial ATA, and by the name, you can hear it's serial. Now in the case of serial, that means one bit at a time, which means that's the bus width, it's one, because they've determined that sending one bit at a time using high speed technology is more efficient than the old traditional drive technologies which work using parallel communication, where we'd have multiple bits that would be sent together, but then they had to be synchronized.
Okay, they determined that serial is faster, so the bus width only matters if you're using an old traditional drive using parallel communications. Next we have IOPS. Now IOPS stands for Input/Output Operations Per Second. Now if you were listening to the words and looking at the letters, you might be thinking that there's an O missing, but there's actually not. It's just one O in IOPS, so the I pretty much is Input/Output together, and then Operations Per Second.
Now this is a value that describes how fast the drive can read and write to the disk. And this I will tell you is one of the major advantages to using SSDs or solid state drives, because you will find that they have much, much higher IOPS. Now we have seek time and latency, and you remember back at RPMs I said keep in mind inverse relationship with latency. All right well we'll get there in just a second. First let's talk about seek time. This has to do once again with only your traditional hard drives that have an actual spindle of disks.
The seek time has to do with how much time it takes for the actuator arms, so there's an arm going back and forth on these disks, how long it takes that arm to get to the correct cylinder on the disk that has the data. Latency is how much time it takes for the disk to spin so that the actuator arm is hovering over the correct sector. As I mentioned earlier, RPMs and latency have this inverse relationship, and it's because the faster the drives are spinning, well obviously the shorter the latency is going to be, or the faster that the arm will get to the location it's supposed to be at.
And the last specification that I want you to be familiar with is understanding the difference between hot swappable drives and non-hot swappable drives. Now hot swappable means that the disk can be added or removed to or from the server without having to power it down. Okay so if you have the ability to add or remove a drive without turning the power off, and without damaging the drive or losing data, then it's hot swappable.
Whereas if you have to actually shut down the server to do so, then it's non-hot swappable. Now it should be fairly obvious that hot swappable technology is going to be a little bit more expensive, but that's something that you're going to have to determine whether you have that need or not. So, these are the seven main disk specifications that you should be familiar with. Now the next thing you have to pay attention to are the interfaces that are available when trying to add a drive to your server.
Now we've talked about some of these. The first one here would be Serial Attached SCSI, or SAS. Now this is a type of SCSI that uses serial operations rather than parallel, which is what the original SCSI used. Now there are some other ways in which it differs. The traditional SCSI bus was what was known as a multi-drop bus, meaning you would attach multiple devices onto a single daisy chain basically, so onto a single cable.
Whereas serial attached SCSI is point to point communication. The other big difference is that traditional SCSI required termination at each end of the cable, whereas again that's not needed with serial attached because it is point to point, you don't need any termination. Now the next and second most common, so these top two are the most common interfaces out there today, would be Serial ATA, which again, using serial communications as opposed to the original ATA specification which was very often known as IDE.
And in today's world, is actually known as P-A-T-A, or PATA, which stands for Parallel ATA. Okay, so I will tell you that the main thing is is that traditionally we started off with parallel communications with your traditional SCSI and IDE, and we've evolved into newer and faster technologies, which both utilize the faster serial communication. Now all that said, I will tell you that there is still traditional SCSI that is out there, especially if you're working with a server that's maybe a little bit older, but still functioning on your network, and so you do need to keep in mind everything that I had already mentioned as far as how traditional SCSI works with parallel.
You need to have termination, because you have a daisy chain, and every device in the SCSI chain requires its own unique SCSI ID. So you could almost think of it kind of like multiple computers on a network, right, on your local area network. They each have to have their own IP address, they have to have their own unique ID. Well the same thing held true with the traditional SCSI communication. Now we also have USB, which probably doesn't come as much of a surprise, right? Because we know that USB can connect pretty much anything.
USB is an amazing communications technology, and storage devices are not an exception. They can be connected via USB. They even have adapters to connect SCSI and USB, but one other thing I want you to know about USB is that some drives, while they connect via USB, they still require their own power source, especially if it's an external hard drive. Whereas other devices now have the ability to utilize the low power output that comes through the USB connection, so then you don't even have to have any power.
You just plug in the drive and you're on your way. Now I will tell you that typically your USB is going to be used maybe not so much for your primary storage, but more what we would refer to as secondary or auxiliary storage. That's where you'll more commonly find USB connectivity. And the last interface which is pretty common in server environments is the Fibre Channel interface. Now this is used to connect devices in a high speed fibre storage network.
Now, there are some specifications to go with this, and I talk about this in a little bit more detail in another video, but basically these networks will use a form of a fibre switch with the devices being connected to the switch using the Fibre Channel interface. There's other terms you may be detail with. HBA which is the host bus adapter. The WWN which is the world wide name, which is a lot like a MAC address where they're all individually recognizable. So again I'll talk about that a little bit more.
We'll talk about HBAs, fabric switches, stuff like that in another video, but I want you to understand that this is another interface that you may work with with your servers, and again it really has to do with mainly when you're working with these high speed storage area networks. Now the one last thing I want to leave you with is I want to make sure that we kind of understand the difference between a traditional hard drive and an SSD, which is a solid state drive. Okay I've mentioned this a few times here in the video, but a traditional hard drive has actual disks.
They spin, you have moving parts. You have an actuator arm that moves back and forth. There's obviously a lot more specifications on a traditional hard drive that you need to pay attention to, things like RPMs, seek time, latency. Whereas your SSD has no moving parts, and therefore tends to be faster, and it also tends to be more reliable especially when it comes to if a drive was dropped, because there are no moving parts, it has a much better chance of surviving.
Whereas a traditional hard drive, things would shift or break. And one other major difference between a traditional hard drive and an SSD is that a traditional hard drive is susceptible to magnetism. You can erase or corrupt the data with a magnet. Whereas an SSD is not susceptible to that same magnetism. All right so that is pretty much the main stuff that you need to know, and once you are familiar with the specifications and you're comfortable with all the specifications, and you know the interfaces that you need, you're ready to go out there and go start purchasing away.
Get that primary storage for your servers going.
- Storage technologies
- Deploying primary storage devices
- Fault tolerance
- Configuring RAID
- Storage capacity planning