RAID Discisions

[pizzaman_314]

I think this forum needs a thread dedicated to discussing the ups pros and cons of raid, where people can ask questions. Comence the discussion

[MegaDave91]

I'm curious as to how many users actually use RAID. My drives are completely independent one another, but I may try RAID 0 some day with different drives.

[HeyKitty]

Not just raid, but all the different type of ways to store data. also, that is this sub-forum

[Senzelian]

Thats the reason the "storage"-forum exists.

 

I'm curious as to how many users actually use RAID. My drives are completely independent one another, but I may try RAID 0 some day with different drives.

I had a RAID 0 with my two 1TB HDDs. Not really worth it =/

[FSBAlston]

Used raid 0 on my desk pc. Upgraded to a ssd a few years later. Haven't looked back. Raid is a good solution for nas storage, but I find myself shying away from it in general consumer use.

[unijab]

 

 

To Raid, or not to Raid?

 

 

Abstract (summary)

 

RAID configurations can be perfect for some situations, but overkill for others. The important thing is to have some kind of solution in place to protect data against disk failure.

 

 

 

 

To Raid, or not to Raid?

Should you protect your data using Raid? It depends on what you're storing

Our recent articles benchmarking various Raid configurations sparked a debate in the Hands On mailbox. Long-term Raid users defended the technique of using multiple hard disks as the only certain way to protect against disk failure. Others believed Raid was overkill for general users and argued that copying files onto an external hard disk was a much better use of resources.

As with all good arguments, there are cases for both sides. Raid can be perfect for some situations, but overkill for others. The important thing is to have some kind of solution in place to protect your data against disk failure. So, with this in mind, we're returning to the subject of Raid, but this time weighing up its pros and cons against simpler solutions to see if it's the right technology for you.

Basics of Raid

In previous issues, we've described in detail how Raid works, so we'll just cover the basics here. Raid stands for Redundant Array of Inexpensive or Independent Hard Disks and, in fact, uses two or more disks to deliver improved performance, reliability or both.

All Raid arrays are managed by a Raid controller, which does the work of distributing the data between the disks and allowing them to be treated by an operating system as a single storage volume. Raid controllers come in many shapes and sizes, and most motherboards now feature one of some description as standard. Alternatively, you can equip a PC with Raid facilities by using an expansion card from companies such as Promise.

Budget Raid controllers provide little more than a number of plugs to connect the hard disks and offload the data management to software running under your PC's main processor. More expensive controllers feature dedicated hardware to handle the disk and data management to deliver much improved performance. Different Raid controllers also support different types of Raid arrays.

The most common type of Raid arrays are known as Raid 0, Raid 1 and Raid 5. Raid 0 reads and writes data on all disks simultaneously, improving speed and making the full capacity available, but at the cost of reliability - if one disk fails, you lose the whole array. Raid 1 makes an identical copy of one disk on another to provide 100 per cent redundancy, albeit at the cost of half the total capacity and a slower speed than Raid 0.

Raid 5 uses three or more disks and delivers 100 per cent redundancy by writing parity data on all of them. The advantage over Raid 1 is losing only one disk's worth of capacity to redundancy, but the calculation of parity data is complex and delivers poor write performance under software-based controllers. Hardware-accelerated Raid 5 controllers solve the performance issues and can deliver read and write speeds approaching Raid 0, but decent models can cost more than £200.

The case for Raid

Since we're talking about protecting your data, we'll dispense with Raid 0 and concentrate on Raid 1 and 5. Once configured by the controller and formatted by your operating system, a Raid 1 or 5 array presents itself as a storage volume, just as a single hard disk would. And like a single hard disk, you can copy files to and from it.

Should everything behave in your system, you won't notice any difference between using a Raid array and a single hard disk, apart from potentially larger capacity and slightly different read and write speeds, depending on the Raid level and the type of controller.

The clever part of Raid kicks in when something goes wrong. Should one of the disks in a Raid 1 or 5 array fail for whatever reason, your data remains fully accessible - and so does any remaining capacity in the array. In fact, as far as how your operating system sees and presents your data, nothing's changed. Life can go on as if nothing happened - except, of course, you now have a failed disk and something needs to be done about it.

The Raid controller, or the software that came with it, will identify the failed disk, allowing you to swap it for a working one. The controller will then rebuild the array, distributing the data onto the new disk. This can take several hours but it happens in the background, so, once again, there's no disruption, other than a temporary reduction in I/O performance, the degree of which again depends on the Raid level and controller type.

If you have hardware that supports hot swappable disks, you can even swap the broken disk for the new one without having to power-down the system.

This 100 per cent uptime is the key benefit of Raid 1 and 5, which are used by PCs that simply have to remain running all the time, such as mission-critical servers. The question is whether you value your data to the same extent.

As we'll explain in the next section, there are definite downsides to Raid, and if you can tolerate some downtime, there are cheaper, simpler and more efficient alternatives.

The case against Raid

The biggest case against Raid is cost. There's the cost of the controller, particularly if you've gone for a hardware-accelerated model, and the cost of the disks in the array. Remember, with Raid 5 you lose one whole disk's worth of capacity, while for Raid 1 you lose half of your total. It can be pretty galling to splash out on, say, two 500GB disks and end up with only 500GB worth of storage.

Then there are the environmental aspects. Having two, three or even more disks running simultaneously consumes proportionately more power, while generating greater heat and noise. This in turn demands better cooling, a bigger case and heftier power supply. There's also the concern that all these disks are gradually wearing themselves out just to protect you against a moment that may not happen for years. But then that's the catch with any type of insurance.

It's also important to understand that a Raid 1 or 5 array with a failed disk, or even a new one that is currently rebuilding, is in a vulnerable state. Should a second disk fail before the array is rebuilt, you're out of luck and will lose all your data - so you should really see it as dodging a bullet and having a window of opportunity to make things right. More sophisticated versions of Raid, such as level 6, can actually tolerate two disks failing, but they're beyond the scope of this column - for now.

The case for another disk

The popular alternative to Raid is to simply back up your data onto a separate hard disk at regular intervals. There are many benefits to this process, the most obvious being that you're able to use every last gigabyte you've paid for.

Since the backup disk only needs to be connected during the backup process, it's also only powered up for short periods, consuming just the power that's required and not wearing itself out unnecessarily.

Most second hard disks used for backup are external models, which brings the additional benefit of portability. After backing up, you can simply move the disk to a different location, greatly reducing the risk of losing everything through fire, flood or theft. You can, of course, also use the disk to transport the data to different locations.

The case against another disk

The big downside to manually backing up onto another disk is you're only covered up to the point you last did it. So if that was, say, one week ago, and your disk fails, then you've effectively lost one week's worth of data.

Another crucial downside is that until the data has been recovered, you can't access it from the failed disk. This can be a big deal for system disks as your PC will, in effect, effect be out of action until the recovery process is completed.

Portability is additionally a double-edged sword, as you can more easily lose or damage an external disk, or accidentally transfer a virus onto it while connected to another system.

Verdict

If you absolutely need 100 per cent uptime and constant protection for your data, Raid is the only way to go. While you'd assume everyone falls into this category, there are many situations where it's actually overkill.

Let's say you back up a media collection of ripped CDs and digital photos onto an external hard disk. Should the original disk fail, it may not matter a great deal if you have to wait a few hours while you fit a new disk and transfer the files across from the external backup. And if you needed the data straightaway, you could access it direct from your external disk.

It's also worth noting that most media collections aren't updated on a constant basis. You only tend to update them after buying a new CD or returning from a trip with a camera full of photos. So having a system that in effect performs a constant backup for data which changes infrequently, as in this instance, is unnecessary.

Where Raid becomes invaluable, though, is for data that is being constantly updated, such as emails and documents. Even if you scheduled a backup every night (which would be very conscientious), you stand to lose up to a day's worth of new files and changes to existing ones. This could be intolerable for many businesses. Raid automatically keeps you protected right up to the moment a disk failure occurs, and allows you to continue working through it as you fit a replacement.

It boils down to the degree of protection you're after for your data. Raid may be necessary for mission-critical data protection, but if the files you're protecting change infrequently and some downtime is acceptable, then backing up onto a single disk represents better overall value. Just remember to perform that backup every time the data changes. For more details on the difference between software and hardware Raid 5, see March's Hands On Performance column at www.pcw.co.uk/2173912. PCW

Word count: 1694

Copyright © VNU Business Publications Ltd. All Rights Reserved.

Title

To Raid, or not to Raid?

Publication title

Personal Computer World

Pages

n/a

Publication year

2007

Publication date

Jul 2007

Year

2007

Publisher

Incisive Media Plc

Place of publication

London

Country of publication

United Kingdom

Publication subject

Computers--Microcomputers

ISSN

01420232

Source type

Trade Journals

[pizzaman_314]

I am using raid 0 on two 840 Evos, and boot-up times are rediculous.  I also get over 1 gig read and write speeds. I love it so far