In the article below, Troy Sniff and Mike McManus, two of our IT Consultants, describe some of the basics around RAID storage inside of a server. Enjoy!
Choosing a RAID Solution
By Troy Sniff + Mike McManus, IT Consultants
Before you can delve into the specifics of choosing a RAID solution, you need to know the basics. Mike explains:
RAID is defined as a “redundant array of independent disks”. All that really means is that you are using multiple hard drives to improve the performance of your server. It also provides redundancy that helps lower your risk of total data loss. The RAID will store the data across many hard drives, but how it handles the data is very different per RAID.
What is the difference between the RAIDs? There are many different types of RAID’s that you can do. Each one has its benefits and down sides. It will all depend on what you are looking for. Each one is different and will depend on if you are looking for higher speeds, lower risk of data loss or a little of both. It will also depend on the demands of your machine. That is where IOPS come into play.
IOPS is defined as “Inputs/Output Operations per Second”. IOPS is basically how many operations are preformed on your hard drive or RAID. It is the amount of times per second the system is reading from or writing to the RAID. You will want to ensure that your RAID can handle the amount of IOPS that is demanded by the server or you will start to get “bottle necks” and that will cause performance issues.
Now that you’ve got some understanding of RAID concepts, Troy can explain some of your options more in depth:
In the server realm, picking the best RAID solution can be a tricky thing. The first thing we need to do is take into consideration the key factors; cost and capacity, performance needs and availability.
There are several flavors of drives these days and the cost is directly related to type and size. For now, let’s focus on SAS and SATA drive types. You can buy SATA drives in the several terabyte ranges for fairly low cost. While SAS capacities are breaking into the terabyte ranges, they tend to cost two to three times that of an SATA drive of the same size. Cost and capacity boils down to the type of RAID you choose. With different types of RAID levels you will lose x number of drives of capacity. An example would be with RAID 5 you would lose one drive of capacity while in RAID 10 you would lose half of your capacity. There are many other details to consider regarding capacity and cost, but this is the first step.
With the different types of drives come differences in performance. Once of the main factors you have to consider is how many IOPS (Inputs/Output Operations per Second) your server or applications are going to require. Once you know this, you can move forward to designing your array layout.
Not only do drives come in different types, they come in different speeds; SATA 7,200 rpm, SATA 10,000 rpm, SAS 10,000 rpm, SAS 15,000 rpm, etc. With the different types and different speeds come different IOPS so we will jump a little more into this.
To calculate IOPS in a drive, you need to take into consideration average latency and average seek time using the formula: 1 / (Avg seek time in sec + Avg latency in sec).|
When talking availability we need to consider a number of things. I won’t go into every single one of them but I will try to highlight some of the important considerations.
The first would be how your selected drives will hold up in a RAID scenario. SAS drives are typically designed for running more hours and larger workloads. They also tend to include the more advanced technologies such as command queuing, reordering, lower bad sector recovery times, etc. There are more examples, but this is the basic idea.
The next thing to consider is how your array will sustain a single drive or multiple drive loss. Let’s take RAID 5 for example. RAID 5 can sustain a single drive loss without data loss but a loss of two drives will result in total failure; no matter the number of drive it consists of. In addition, the more drives you add to it, the higher chance you have to run into a total loss scenario. RAID 10 can sustain multiple drive losses. The number of possible drive loss in RAID 10 depends on the total number of drives in the array. For example, a four drive RAID 10 can lose two drives while a six drive RAID 10 can lose 3. Now please note, the numbers of drive losses must happen in a specific scenario. Or in other words, you have to lose specific drives in specific parts of the array.
The last factor to consider is the rebuild times when you do have a drive failure. The different levels of RAID all have different times for rebuild but the important thing to think about is exposure during rebuild. During the rebuild, you are not only in a degraded state as far as performance but you are also in a very vulnerable state. What happens if you lose another drive during this time? It could mean total failure, and if not, this means an even more degraded state with an even higher chance of total failure. Taking that into account, it will make you think very hard to which RAID to choose.
In conclusion, it should be mentioned that there are many things to consider that haven’t been covered here. Choosing a RAID solution can be overwhelming, but Greystone is here to help. Ask your consultant to help explain the process, or let us know whenever you have questions! Even email me if you have any questions, comments or even complaints: firstname.lastname@example.org.