60GB IDE Hard Drive 9GB SCSI Hard Drive 60GB IDE Hard Drive 9GB SCSI Hard Drive

The above are some the hard drive options listed at Nocster. what is the best to get? IDE or SCSI. Isn't SCSI much larger and therefore better?
Bradley

Larger? The IDE hard drive is 60 GB and the SCSI hard drive is 9 GB. If you don't need more than 5 GB to 7 GB of disk space, then get the SCSI hard drive for better performance. Otherwise, get the IDE hard drive.

It also depends on what kind of motherboard is in use. If its a cheapy motherboard with a cheapo IDE controller on it, I'd either get a better controller installed for IDE, or go SCSI. We've had enough motherboard replacements due to the IDE controllers flaking on us that we are replacing them with SCSI drives when failure occurs.

if you can afford it, i'd go with SCSI, although you will first want to work out how much disk space you will really need. if you don't need 40GB, then go with SCSI.

SCSI drive are better for high I/O applications, for example, mail, web, news server etc. Also depends from the load and if it is low, an IDE drive will do good (hopefully it's ATA100 with 7.2k RPM). Also, I have seen info that new ATA100 is doing very well comparing with the old SCSI 2.

SCSI (U160) has 160Mbps transfer rate, I dont think IDE can deliver it...

ps. there are some good IO test software you can run on Linux, for example.

I agree with the Motherboard comment. If you motherboard can support Ultra ATA/100 then go with an IDE/ATA drive. Otherwise go with SCSI.

ATA/100, as the name suggests, can transfer data to the motherboard at UP TO 100 megabytes per second. My desktop here at home has an ATA/66, (which, theoretically, should get 66 megabytes per second), but I never got any higher than 50 with benchmarks... :(

Also remember that the larger the drive, the less efficient it is. It takes longer for it to scan to a different portion on the disk. 120Gig disks aren't always better than 40 gigs. :)

As for SCSI... It is MUCH faster and has a lot more umph, and will last longer with more reliability, but, at a much higher price.

You might also want to look at whether or not they offer RAID... Even RAID 1 is better than nothing (zero never did anything for me... just wasn't good enough), but 5 is best.

Personally, I'd go for a 9 or 18 Gigabyte SCSI with RAID-5... :) But tha't just screwed up old me. :)

- James

the best RAID configuration depends from the application. For some it will work with RAID-5 but for some RAID-1,RAID-0,RAID-10,RAID-4 will be better.

Consider RAID-0 (or RAID-10 for redundancy) or RAID-1 for example for DB servers....
For web servers RAID-5 is the best option..

PS. For RAID-5 you will need a minimum of 3 hard drives!

If you don't need more than 9 GB including the OS, log files, web site files, etc, the SCSI drive will be noticably faster and the one to go with for a web server.

If you might need more space within the forseeable future, you should go with the IDE drive.

i have a few newbie questions about RAID technology:

what are the MAIN BENEFITS of using RAID?

i only know one reason at the moment ie. doing backups. but could someone explain the benefits of doing backups with RAID?

does installing RAID 'software' on a ded server cost extra money?

any good links on the net where i can pick up some more info? i did a prelim search on google, but couldn't get good links.

thanks to all in advance.

Hi Chang,

While most modern operating systems will do software raid (included in ther OS) the only way to go is hardware RAID. This means that the SCSI controller on the motherboard/seperate card, has integrated software which controls multiple hard-drives in a raid array. Manufacturers ship boot disks/cd's which allow you to boot the machine and configure which RAID option you want and how you want to set things up before you install the operating system.

Normal uses are for extended storage or redundancy. With extended storage, the drives combine to appear as a single large drive to your OS, with redundancy, data is mirrored across two or more drives which allows for a hard drive failure to occur without causing your machine a problem.

HTH,
Steve

raid technology is used for the following benefits:

-better read (RAID-0, RAID-1, RAID-5, RAID-10)
-better write (RAID-0, RAID-1, RAID-10)
-redundancy (except RAID-0)

There are also RAID-2, RAID-3, RAID-4 but they are not so common - thought some people use these technologies for very specific applications - sometimes, DB.

ps. raid-5 is the most cost-effective, but sometimes the cost is not the factor (large scale IO applications - mail, db..)

Most used:

RAID 0 :

Files are separated and sent to 2 HD's. So if you write a file of 20MB the file will be half on disk A and half on disk B. You won't see that in the OS cause these actions are done by the controller, and in the OS the files are automatically parsed together. In theory this will give you the double read/write speed of your box.

RAID 1: Mirror

Files stored on disk A will be automatically stored on disk B.
Do, if one HD fails, all files will be available from the other disk.
This type of raid offers a redundant storage system.



Note that this is just a simple explanation to understand what they do, in fact it's all very complex

WOW. And I spent one and a half hours searching and accessing links via google and couldn't understand these simple things... and when I read your replies here... it all becomes clear as daylight!

WHT RULES !!! :)

Keep posting more interesting generic knowledge about RAID if there is anything not already mentioned here guys!

You're all awesome! As always!

I want to Jam 4-6 60Gig EIDE harddrives into 1 server. Since the servers is quite large I want to use it for backing up files of two other webservers. I want to rsync data every night at 12 AM from the two boxes. I also want to host with my large server. Should I do some sort of RAID setup? Which one or how would you guys do it?

Originally posted by ClusterMania
I want to Jam 4-6 60Gig EIDE harddrives into 1 server. Since the servers is quite large I want to use it for backing up files of two other webservers. I want to rsync data every night at 12 AM from the two boxes. I also want to host with my large server. Should I do some sort of RAID setup? Which one or how would you guys do it?

It maybe better - and cheaper - to separate the big server if you have room into two servers. It would be cheap to to buy a 1U with 3 drives and put 3 IDE 120s or 160s at 5400. This is how we do server back ups. We do not run RAID on our backup servers. When all said and done, we will be backing up almost 40 servers with each 1U. We got this server at QSOL from Joe for about $2000.

Many will argue that RAID5 storage is worth the extra dollors, but in our experience, cheaper with more storage space is better.

Also, you said you want one big server for back ups and hosting. Remember, back ups are best if they are not on the server you will also be backing up. If someone hacks it for example, you will wish you had your data backed up elsewhere. Of course you can use one of your other servers for that.

I already have a IBM netfinity 7000 M10 sitting around doing nothing. It's quite huge so I was thinking of ways I could put it to use.

I have a dumb question:

Do most normal hosts have RAID 1 harddrive installed on the server? Is that how they do backups so that if something bad happens the data isn`t lost?

Like the servers at rackshack don`t come with RAID right...and that`s the reason why i can`t backup my customers sites anywhere when i have a server from there for example?

:confused:

Originally posted by Frosty
Do most normal hosts have RAID 1 harddrive installed on the server? Is that how they do backups so that if something bad happens the data isn`t lost?

No. Simply said, RAID is no backup. With a RAID configuration the data is distributed over two or more drives. Most failures happen on the software or user side, so any software or user destroying data will immediately destroy the data on all drives.
A backup (tape drive, network backup, second independent drive) is done e.g. daily and copies the data to an independent target. If something happens you are able to get the data from there...