May 9, 2008
Dan Ushman

Every once and a while someone asks me what he or she should choose to protect their data. There are a lot of choices: half dozen types of RAID arrays, second-disk-backups, off-site backups, network storage, and more.

Before I discuss the various options, I want to take a moment to discuss the various levels of RAID Arrays that are available and their individual advantages and disadvantages.

RAID stands for Redundant Array of [Independent or Inexpensive] Disks. That means that RAID uses various configurations of hard drive to avoid damage in the event that one (or sometimes more) of the drives fail.

There are numerous types of RAID Arrays. At SingleHop we offer four of the most common options.

These are:


RAID-0

This level of RAID offers very little data redundancy protection and should never be used alone. If one of the disk drives fails, the entire array will be gone. However, the shortfalls in redundancy are balanced out by a tremendous advantage in terms of performance. In a RAID-0 configuration data is stripped onto two drives or more drives, but for purposes of this example I am going to assume that only two drives are being used. Part of the data is on drive A and part of the data is on drive B. What happens, in simplified terms, is the data is split into chunks and written to the same position on the two disks. Then, when read, the disks both return part of the data to the RAID controller and it combines them into the file you are looking for.A good way to remember this is this: The 0 in RAID is the amount of data you will have right if any of the hard drives in your array fail.

RAID-1

This is one of the most common RAID levels. It is very similar to RAID-0 except that the data is not striped, but instead an identical write is made to both disks. It also offers some performance options, but not as much of a performance increase as RAID-0 offers.

RAID-5

One of our favorite types of RAID is RAID-5. For a RAID-5 configuration to work, you need a minimum of three hard drives. We typically recommend four. The data is stripped across the drives with a portion of the data on various drives, however in the event one of the drives fails it is typically recoverable by swapping out the drive and rebuilding the array. Further, RAID-5 offers some performance increases. However, in the event of multi-drive failure there is an unlikely chance that the array can be rebuilt. We typically recommend that RAID-5 be used along side with a separate backup drive or separate backups on a second machine or our backup storage servers.

RAID-10, also known as RAID-1+0 sometimes RAID-0+1 (No Technicalities Please)

RAID-10 is a combination of RAID-0 and RAID-1. Basically, think of a RAID-0 array with two RAID-1 arrays in it. This type of configuration requires an even number of hard drives (4 or more) that are on two separate arrays, linked by a third array. This offers both performance and redundancy at the cost of increased complexity. Increased complexity is not a big deal if you have a team of experience system administrators to manage your managed dedicated server.

While this post covers the most common types of RAID configurations, there are other options such as RAID-3 or RAID-6, which have their advantages and disadvantages. These two types of arrays are similar to other arrays and are typically used only in special situations.

Now, moving on to backups. There are several different types of backups. Actually, there are a lot of different types of backups.

Same-server, different drive backups.
This is very common at a lot of hosting companies. They install a secondary hard drive to your server and configure the server to copy the data to it once per night, or incrementally throughout the day. The advantage is that it is fairly simple to manage, and restoring a backup is faster because it is all in the same machine. However, if something happens to that machine or the second drive goes bad, you may be out of luck.

Different server incremental or nightly backups.
Having a separate backup server is always a good idea. Having a primary server with RAID and a second backup server with RAID also offers a great deal of security and redundancy, especially if you are using RAID-1 or RAID-5.

Tape backups.
Tape backups are the de-fact-o standard at a lot of organizations. However, they have a number of disadvantages. First, in most cases tape backups are stored off-site. Someone has to change the tapes, then transport them to a second secure storage facility. This is very similar to making a Xerox of a document and putting it in a safe deposit box at a bank. The down side is, well, there are a lot of them: You have to transport them. You have to store them. You have to rotate them frequently. There is a lot of manual work involved and automated systems are expensive and also require a great deal of man time.

MySQL or Database Replication
This is a technology that we use at SingleHop a great deal. Basically, in a database server configuration, a second server is setup to mirror the first in real time. If both machines also have redundant RAID (anything except RAID-0) then there is a great deal of security.

Internally, we use a combination of RAID-5, MySQL replication and secondary incremental and nightly backups. Your own needs will vary and if you are not sure you should always speak to a trained and experienced system administrator before deciding what you intend to do.

None of these solutions carry a huge expense on their own. However, the pricing can add up to be prohibitive if you are very paranoid. In most situations a simple RAID-1 array provides enough redundancy to sleep better. I guess the best moral to the story would be not to put all your eggs in one basket. If your data is valuable, if you will go out of business by losing it, then RAID + second server backups are well worth the additional cost.

Comments

    This is a great post explaining how Raid works and can benefit any server.

    Posted by dsalcedo on May 18, 2009 Reply

Leave a Comment