Raid

Hey I've been hearing a lot of talk about RAID and was wondering what it was exactly, and if it's possible to upgrade a laptop that doesn't have it so that it does.

 

Hi Wohh,

Help is here! I know all about RAID.

What is RAID? Well, RAID is known as a redundant array of independent disks. It is a special way of storing the same data in different places on multiple hard disks.

What happens is that by placing data on multiple disks, I/O (i.e. input/output) disk operations can overlap in a balanced way. Thus, RAID improves performance.

Now, since you can also have multiple disks in a RAID setup, you also increase the mean time between failures (MTBF) of your hard disks. This means that RAID also increases fault tolerance - by storing data redundantly.

When you set up a RAID array of hard disks, the operating system sees the array as a single logical hard disk. What happens is that RAID employs a technique called disk striping, which is a fancy way of partitioning each drive's storage space into small little units. These little units or stripes can be retrieved by the operating system easily since they contain individual addresses.

To explain this in more detail, there are (at least) nine types of RAID plus a non-redundant array (RAID-0):

RAID-0: This base technique performs striping but has no redundancy of data. RAID-0 offers the best performance but no fault-tolerance.

RAID-1: We call this type of RAID array 'disk mirroring'. It simply consists of at least two drives that duplicate the storage of data. The data is not striped. So what happens is that you get improved read performance - either disk can be read by the operating system at the same time. Write performance remains the same as for in single disk. RAID-1 provides the best performance and the best fault-tolerance.

RAID-2: In RAID-2, you get striping of data across disks with some error checking and correcting (ECC) information. It is very similar to RAID-3 (RAID-3 has some extra features).

RAID-3: In this set up, you get data striping and also a dedicated drive to storing error checking and parity information. You find that the embedded error checking (ECC) information is typically used to detect errors. RAID-3 is very good for systems with long record applications.

RAID-4: RAID-4 uses large data stripes. What this means is that you you can read records from any single drive. You get to take advantage of overlapped I/O for read operations. However, write operations have to update the error checking drive - so no I/O overlapping is possible. In essence, RAID-4 has no advantage over RAID-5.

RAID-5: RAID-5 is a setup where you have a 'rotating parity array'. So you address the write limitation in RAID-4. What this means is that you get all read and write operations being overlapped. You need about three to five disks for the array. It's best for systems in which performance is not critical or where only a few write operations are needed.

RAID-6: This array of disks is very similar to RAID-5. It includes a a second parity scheme that is distributed across different drives. What you get is a setup that is extremely high fault and drive-failure tolerant.

RAID-7: RAID-7 is not so common, but it includes a real-time embedded operating system as a controller! So what you have is a high-speed bus, and essentially a stand-alone computer that monitors your hard disks.

RAID-10: In RAID-10, we combine RAID-0 and RAID-1 and try to get the best of both worlds - speed and fault tolerance. Of course, this comes at a much higher cost.

We can identify two subtypes: In RAID-0+1, your data is organized as stripes across multiple disks. The striped disk sets are then mirrored. In RAID-1+0, your data is mirrored and these mirrors are striped.

RAID-50 (or RAID-5+0): Again pretty rare, but this set up consists of a series of RAID-5 groups and striped in RAID-0 fashion. You get an end result that improves RAID-5 performance but does not reduce its data protection capabilities.

RAID-53 (or RAID-5+3): In this setup, we use striping (like in RAID-0) for the virtual disk blocks you find in RAID-3. What we get is higher performance than RAID-3, but (of course) at much higher cost.

RAID-S (or Parity RAID): This is a proprietary method for striped parity RAID. It was invented by EMC Symmetrix but is no longer in use. Functionally, it is similar to RAID-5 with some performance enhancements.


Whew - sorry for going overboard with the technical terms and specifics. Hope that helps you out. RAID is a pretty confusing topic - and I think that most laptop owners don't use RAID (unless you're a hardcore nut). I find RAID arrays more prevalent in desktop systems. If you want to find out more, drop me a private message, I can point you to some good resources / books about RAID.

 

unless your notebook is one of the very very few that can run multiple hard drives, then no, you can not enable RAID in your machine.

 

Inexpensive disks is what it stands for. (some marketing people have tried to get it changed to independent, but the 'I' originally stood for Inexpensive)

And RAID 0 is far from always a good choice for performance. For most tasks, it's within 1-3% of the performance you get with a single disk.

And nope, you most likely can't run RAID on your laptop. The entire point of it is to use multiple harddrives together. Which obviously requires you to have multiple harddrives.

 

Inexpensive... Independent... there's a lot of debate on the topic. Well, not a lot of debate, because nobody really cares all that much, but I've seen no evidence that one came before the other. Best I can tell is the mistake was made and perpetuated very early on, so they're basically interchangeable.

Regarding RAID on your laptop, like everyone said, only if your laptop supports two drives. Most that have two truly internal drives these days will probably support fake RAID. That is hardware supported software RAID. If not, you can still use software RAID if your operating system supports it. This is also true if you have an option to run a drive in a modular bay like on some HP and Lenovo business class laptops, for example.

If you're looking for redundancy you could also run external drives in software RAID 1 or 5, but 0 would be pretty pointless considering the speed delta between IDE/SATA and USB/Firewire. But, if you just want redundancy you'd be better served by a NAS.

 

Read the original paper from UCB, and then tell me there's no evidence please. I have. It said inexpensive.

There's nothing to debate unless you are debating for the sake of debating.

 

Kudos to you for knowing the true origination. But honestly, the acronym has degraded to the point where the two are interchangeable, and neither one or the other can really be considered correct. When you get right down to it, yes, but try telling that to the thousands of people who say 'independent.'

Like I inadequately tried to express, it's not so much that there's debate on the topic, it's just that people use both. It doesn't matter, and RAID in practice has probably gone well beyond the original intents that warranted the use of the term 'inexpensive.'

 

Oh ok. Thanks for all the info, helped a lot.