To RAID or not to RAID? -that is the question...

Hello all, I am trying to set up an Informix database (7.3) on Linux (Redhat 6.0) on an Intel box (dual PII 400). Since this is going to be a production machine, I will be using mirroring of database chunks. I want to get some advice from people who have setup or have considered setting up such a configuration as regards to disk storage: Will I be better off using RAID or should I use dual SCSI controllers and just mirror the database chunks. Does it make sense to mirror the database chunks on a RAID? Lets assume that cost is a not a factor in the decision -I just want to get the best possible setup. I have been considering the Mylex 960 DAC for the RAID and Adaptec 2940U2W for the SCSI controllers. Thanks in advance for your time. Sandeep Sent via Deja.com http://www.deja.com/ Share what you know. Learn what you don't.

In article <7lmaat$gs3$1@nnrp1.deja.com>, sandeep.singh@usa.net says... > >Hello all, > >I am trying to set up an Informix database (7.3) on >Linux (Redhat 6.0) on an Intel box (dual PII 400). >Since this is going to be a production machine, I >will be using mirroring of database chunks. I want >to get some advice from people who have setup or >have considered setting up such a configuration as >regards to disk storage: Will I be better off using >RAID or should I use dual SCSI controllers and just >mirror the database chunks. Does it make sense to >mirror the database chunks on a RAID? > >Lets assume that cost is a not a factor in the decision >-I just want to get the best possible setup. I have been >considering the Mylex 960 DAC for the RAID and Adaptec >2940U2W for the SCSI controllers. Thanks in advance for >your time. > >Sandeep > > > > host independant RAID 5 may turn out to be a better option "mirroring" is not always the best RAID option and RAID 5 is just as secure AND sometimes faster (certainly cheaper) than RAID 1 (mirroring) check out the cmd (www.cmd.com) & chapparral (www.chaparraltech.com) sites _____ . . ' \\\\ . . |>> O// . . | \\_\\ . . | | | . mailto:highend@srmc.com . . . | / | . (800)324-6711 / (800)720-7618 fax . . . | / .| http://sweb.srmc.com/andys/index.html . . | / | http://kalypso.cybercom.net/scsiper/andys.html ...o | ---------------------------------------------------------------------- Authorized - DIRECT VAR/VAD/Distributor for new SCSI/FC-AL peripherals from: IBM, Seagate, Quantum, Adaptec, Sony, Yamaha, ATTO, JNI, ect

Now here's a guy I'll have to hunt down! Art S. Kagel highend wrote: > > In article <7lmaat$gs3$1@nnrp1.deja.com>, sandeep.singh@usa.net says... > > > >Hello all, > > > >I am trying to set up an Informix database (7.3) on > >Linux (Redhat 6.0) on an Intel box (dual PII 400). > >Since this is going to be a production machine, I > >will be using mirroring of database chunks. I want > >to get some advice from people who have setup or > >have considered setting up such a configuration as > >regards to disk storage: Will I be better off using > >RAID or should I use dual SCSI controllers and just > >mirror the database chunks. Does it make sense to > >mirror the database chunks on a RAID? > > > >Lets assume that cost is a not a factor in the decision > >-I just want to get the best possible setup. I have been > >considering the Mylex 960 DAC for the RAID and Adaptec > >2940U2W for the SCSI controllers. Thanks in advance for > >your time. > > > >Sandeep > > > > > > > > > > host independant RAID 5 may turn out to be a better option > "mirroring" is not always the best RAID option > > and RAID 5 is just as secure AND sometimes faster > (certainly cheaper) than RAID 1 (mirroring) > > check out the cmd (www.cmd.com) & > chapparral (www.chaparraltech.com) sites > > _____ . . > ' \\\\ . . |>> > O// . . | > \\_\\ . . | > | | . mailto:highend@srmc.com . . . | > / | . (800)324-6711 / (800)720-7618 fax . . . | > / .| http://sweb.srmc.com/andys/index.html . . | > / | http://kalypso.cybercom.net/scsiper/andys.html ...o | > ---------------------------------------------------------------------- > Authorized - DIRECT VAR/VAD/Distributor for new SCSI/FC-AL peripherals > from: IBM, Seagate, Quantum, Adaptec, Sony, Yamaha, ATTO, JNI, ect

I would tend to agree, move it off the machine onto a dedicated raid system the CMD's are very nice units. I saw one go on ebay this week for $660 it was filled with controllers + quite a bit of ram http://cgi.ebay.com/aw-cgi/eBayISAPI.dll?ViewItem&item=124037270 Tony Platt highend wrote in message <7lt7od$ho5$1@birch.prod.itd.earthlink.net>... >In article <7lmaat$gs3$1@nnrp1.deja.com>, sandeep.singh@usa.net says... >> >>Hello all, >> >>I am trying to set up an Informix database (7.3) on >>Linux (Redhat 6.0) on an Intel box (dual PII 400). >>Since this is going to be a production machine, I >>will be using mirroring of database chunks. I want >>to get some advice from people who have setup or >>have considered setting up such a configuration as >>regards to disk storage: Will I be better off using >>RAID or should I use dual SCSI controllers and just >>mirror the database chunks. Does it make sense to >>mirror the database chunks on a RAID? >> >>Lets assume that cost is a not a factor in the decision >>-I just want to get the best possible setup. I have been >>considering the Mylex 960 DAC for the RAID and Adaptec >>2940U2W for the SCSI controllers. Thanks in advance for >>your time. >> >>Sandeep >> >> >> >> > >host independant RAID 5 may turn out to be a better option >"mirroring" is not always the best RAID option > >and RAID 5 is just as secure AND sometimes faster >(certainly cheaper) than RAID 1 (mirroring) > >check out the cmd (www.cmd.com) & >chapparral (www.chaparraltech.com) sites > > > _____ . . >' \\\\ . . |>> > O// . . | > \\_\\ . . | > | | . mailto:highend@srmc.com . . . | > / | . (800)324-6711 / (800)720-7618 fax . . . | > / .| http://sweb.srmc.com/andys/index.html . . | >/ | http://kalypso.cybercom.net/scsiper/andys.html ...o | >---------------------------------------------------------------------- >Authorized - DIRECT VAR/VAD/Distributor for new SCSI/FC-AL peripherals >from: IBM, Seagate, Quantum, Adaptec, Sony, Yamaha, ATTO, JNI, ect > >

In article <7lmaat$gs3$1@nnrp1.deja.com>, There are a couple of factors to consider when using RAID. 1. How much fault tolerance do you want. 2. What type of performance are you looking for. A RAID 5 setup is tolerant to a point of a single failure. Even if you have a global hot-spare to auto rebuild, you still run the chance of losing your data "if" you have a second drive failure during a rebuild. RAID 5 performance is good with reads. Write performance varies depending on the amount of write-back cache installed on the controller. A RAID 10 (striped pairs) is tolerant of multiple drive failures. The down side is that is requires double the drives for the required storage capacity. RAID 10 performance is good with reads and writes. Since you are using a database, Ultra2 SCSI would be the way to go. Currently, SCSI RAID controllers tend to do better with random small block I/O than a Fibre Channel Solution. We have both solutions. Another way to get the best performance would be to set you RAID volumes up where your temp files run in RAID0 (best read/write performance no fault tolerance) and your main data as either RAID 5 or 10. All of these solutions are "hardware-based". Dan Folwell Sarvannah "Fault Tolerant Solutions For Open Computing Environments" http://www.sarvannah.com 770-928-9236 Fax 678-445-0352 sandeep.singh@usa.net wrote: > Hello all, > > I am trying to set up an Informix database (7.3) on > Linux (Redhat 6.0) on an Intel box (dual PII 400). > Since this is going to be a production machine, I > will be using mirroring of database chunks. I want > to get some advice from people who have setup or > have considered setting up such a configuration as > regards to disk storage: Will I be better off using > RAID or should I use dual SCSI controllers and just > mirror the database chunks. Does it make sense to > mirror the database chunks on a RAID? > > Sent via Deja.com http://www.deja.com/ Share what you know. Learn what you don't.

Thanks y'all for your comments/suggestions. I guess I must be missing something cause I did not get what a lot of people talked about using RAID5. RAID5 will no doubt be faster but then there is no redundancy. Maybe two RAID5 channels and then database mirroring across the two controllers? Will someone please comment on this: Suppose I have a choice of having one single RAID1/0 (or even just RAID1) controller OR having two Ultra2 SCSI channels and then use Informix mirroring across the two cont- rollers. Which one makes more sense? Here is my configuration: Dual PII 400/1GB, RedHat 6.0, Informix IDS 7.3. Thanks again... Sandeep In article <37823220.F5D3079F@bloomberg.net>, kagel@bloomberg.net wrote: > Now here's a guy I'll have to hunt down! > > Art S. Kagel > > highend wrote: > > > > In article <7lmaat$gs3$1@nnrp1.deja.com>, sandeep.singh@usa.net says... > > > > > >Hello all, > > > > > >I am trying to set up an Informix database (7.3) on > > >Linux (Redhat 6.0) on an Intel box (dual PII 400). > > >Since this is going to be a production machine, I > > >will be using mirroring of database chunks. I want > > >to get some advice from people who have setup or > > >have considered setting up such a configuration as > > >regards to disk storage: Will I be better off using > > >RAID or should I use dual SCSI controllers and just > > >mirror the database chunks. Does it make sense to > > >mirror the database chunks on a RAID? > > > > > >Lets assume that cost is a not a factor in the decision > > >-I just want to get the best possible setup. I have been > > >considering the Mylex 960 DAC for the RAID and Adaptec > > >2940U2W for the SCSI controllers. Thanks in advance for > > >your time. > > > > > >Sandeep > > > > > > > > > > > > > > > > host independant RAID 5 may turn out to be a better option > > "mirroring" is not always the best RAID option > > > > and RAID 5 is just as secure AND sometimes faster > > (certainly cheaper) than RAID 1 (mirroring) > > > > check out the cmd (www.cmd.com) & > > chapparral (www.chaparraltech.com) sites > > > > _____ . . > > ' \\\\ . . |>> > > O// . . | > > \\_\\ . . | > > | | . mailto:highend@srmc.com . . . | > > / | . (800)324-6711 / (800)720-7618 fax . . . | > > / .| http://sweb.srmc.com/andys/index.html . . | > > / | http://kalypso.cybercom.net/scsiper/andys.html ...o | > > ---------------------------------------------------------------------- > > Authorized - DIRECT VAR/VAD/Distributor for new SCSI/FC-AL peripherals > > from: IBM, Seagate, Quantum, Adaptec, Sony, Yamaha, ATTO, JNI, ect > Sent via Deja.com http://www.deja.com/ Share what you know. Learn what you don't.

sandeep.singh@usa.net wrote: : Thanks y'all for your comments/suggestions. I guess : I must be missing something cause I did not get what : a lot of people talked about using RAID5. RAID5 will : no doubt be faster but then there is no redundancy. : Maybe two RAID5 channels and then database mirroring : across the two controllers? : : Will someone please comment on this: Suppose I have : a choice of having one single RAID1/0 (or even just : RAID1) controller OR having two Ultra2 SCSI channels : and then use Informix mirroring across the two cont- : rollers. Which one makes more sense? Raid 5 is mirrored. There is enough data mirrored on the disks so if any one drive goes down you can rebuild the drive completely from the remaining drives. You get the speed of striping plus the fault tolerance of mirroring. Of course you only get 1/2 of the space the drives have to offer. The only downside is you have to have at least 3 drives. Regards, Salem

Salem Lee Ganzhorn wrote: > > sandeep.singh@usa.net wrote: > : Thanks y'all for your comments/suggestions. I guess > : I must be missing something cause I did not get what > : a lot of people talked about using RAID5. RAID5 will > : no doubt be faster but then there is no redundancy. > : Maybe two RAID5 channels and then database mirroring > : across the two controllers? > : > : Will someone please comment on this: Suppose I have > : a choice of having one single RAID1/0 (or even just > : RAID1) controller OR having two Ultra2 SCSI channels > : and then use Informix mirroring across the two cont- > : rollers. Which one makes more sense? > > Raid 5 is mirrored. There is enough data mirrored on the disks so if any > one drive goes down you can rebuild the drive completely from the remaining > drives. You get the speed of striping plus the fault tolerance of mirroring. > > Of course you only get 1/2 of the space the drives have to offer. > > The only downside is you have to have at least 3 drives. > Regards, > Salem You may want to research this yourself: http://www.acnc.com/

On 7 Jul 1999 00:44:21 GMT, Salem Lee Ganzhorn <slganzho@unity.ncsu.edu> wrote: |sandeep.singh@usa.net wrote: |: Thanks y'all for your comments/suggestions. I guess |: I must be missing something cause I did not get what |: a lot of people talked about using RAID5. RAID5 will |: no doubt be faster but then there is no redundancy. |: Maybe two RAID5 channels and then database mirroring |: across the two controllers? |: |: Will someone please comment on this: Suppose I have |: a choice of having one single RAID1/0 (or even just |: RAID1) controller OR having two Ultra2 SCSI channels |: and then use Informix mirroring across the two cont- |: rollers. Which one makes more sense? | |Raid 5 is mirrored. There is enough data mirrored on the disks so if any |one drive goes down you can rebuild the drive completely from the remaining |drives. You get the speed of striping plus the fault tolerance of mirroring. | |Of course you only get 1/2 of the space the drives have to offer. | |The only downside is you have to have at least 3 drives. The only RAID varieties in common use are: 0 Plain striping, no redundancy. + Faster reading and writing ( spread over all drives ) - Loose 1 drive = loose all the data. 1 Mirroring. Most redundant, but needs 2x the drive space. + Faster reads ( can read from either drive ) + Can loose a drive and still keep going. - Writes are slightly slower, data has to be written twice. 5 Needs 3+ drives. Data is striped acros the drives, but 1 drive worth of data is made up of checksums of the other data. This is spread out over all the drives. + Faster reads, spread out over all drives, + Can loose a drive and keep going ( with reduced speed ) - Slower writes. Need to recompute and wite the checksum as well A combination of Striping and Mirroring is sometimes called RAID 1+0, RAID 0+1, or RAID 10 -- Reverend Paul Colquhoun, postmaster@andor.dropbear.id.au Universal Life Church http://andor.dropbear.id.au/~paulcol -=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=- xenaphobia: The fear of being beaten to a pulp by a leather-clad, New Zealand woman.

Paul Colquhoun (postmaster@andor.dropbear.id.au) wrote: : On 7 Jul 1999 00:44:21 GMT, Salem Lee Ganzhorn <slganzho@unity.ncsu.edu> wrote: : | : |Raid 5 is mirrored. There is enough data mirrored on the disks so if any : |one drive goes down you can rebuild the drive completely from the remaining : |drives. You get the speed of striping plus the fault tolerance of mirroring. : | : |Of course you only get 1/2 of the space the drives have to offer. : | : |The only downside is you have to have at least 3 drives. : : : The only RAID varieties in common use are: : : 0 Plain striping, no redundancy. : + Faster reading and writing ( spread over all drives ) : - Loose 1 drive = loose all the data. : : 1 Mirroring. Most redundant, but needs 2x the drive space. : + Faster reads ( can read from either drive ) : + Can loose a drive and still keep going. : - Writes are slightly slower, data has to be written twice. : : 5 Needs 3+ drives. Data is striped acros the drives, but 1 drive : worth of data is made up of checksums of the other data. This : is spread out over all the drives. : + Faster reads, spread out over all drives, : + Can loose a drive and keep going ( with reduced speed ) : - Slower writes. Need to recompute and wite the checksum as well : : A combination of Striping and Mirroring is sometimes called : RAID 1+0, RAID 0+1, or RAID 10 I appologize, I was thinking of 1+0. Regards, -- Salem Lee Ganzhorn... slganzho@unity.ncsu.edu

Not even close :) Raid 5 is NOT mirroring. Raid 5 does an XOR of the data spread across multiple drives and stores that result on the parity drive. The parity is usually calculated at the block level and the data is stripped across the drives in the Raid set. When a data drive fails, the data on that drive or volume can be 'regenerated' by XORing the parity data with the data on the surviving devices. As to the original posters comment that Raid 5 will be faster, this is not necessarily true and in most cases Raid-5 will be slower than Raid-1 (mirroring). This is especially true in when there is a lot of write activity. The reason is that when data is written to a Raid-5 volume, the parity for that Raid rank must also be read, recalculated and re-written. This may also involve reading the other data volumes in order to do the XOR calculation. Mirroring on the other hand only requires 2 writes and if you use a caching Raid controller this can be very speedy since the writes go into cache and are later destaged to disk when the disk is less busy. There are other factors to consider as well such as adding drives, increasing the size of the Raid group and so on. Raid-5 is usually much more difficult to add-on to. You usually have to back-up your data, rebuild the raid-5 group and then relay the data back onto disk, which btw usually requires a full rebuild of the Raid-5 parity information. Good luck Cheers TC Salem Lee Ganzhorn wrote in message <7lu7t5$q3l$2@uni00nw.unity.ncsu.edu>... >sandeep.singh@usa.net wrote: >: Thanks y'all for your comments/suggestions. I guess >: I must be missing something cause I did not get what >: a lot of people talked about using RAID5. RAID5 will >: no doubt be faster but then there is no redundancy. >: Maybe two RAID5 channels and then database mirroring >: across the two controllers? >: >: Will someone please comment on this: Suppose I have >: a choice of having one single RAID1/0 (or even just >: RAID1) controller OR having two Ultra2 SCSI channels >: and then use Informix mirroring across the two cont- >: rollers. Which one makes more sense? > >Raid 5 is mirrored. There is enough data mirrored on the disks so if any >one drive goes down you can rebuild the drive completely from the remaining >drives. You get the speed of striping plus the fault tolerance of mirroring. > >Of course you only get 1/2 of the space the drives have to offer. > >The only downside is you have to have at least 3 drives. >Regards, >Salem >

Tony C wrote in message <931333286.482062@cacheraq001>... >Not even close :) well it helps if you give the RIGHT advice >Raid 5 is NOT mirroring. Raid 5 does an XOR of the data spread across >multiple drives and stores that result on the parity drive. The parity is >usually calculated at the block level and the data is stripped across the >drives in the Raid set. When a data drive fails, the data on that drive or >volume can be 'regenerated' by XORing the parity data with the data on the >surviving devices. Wrong, you are thinking of DATA guarding (RAID 4) Raid 5 Distributed Data Guarding stripes the parity data across ALL disks in the set (like the name says) Raid 5 offers Best protection Increased throughput Most cost effective Striping parity data generates redundant information that can be used for the following Detecting errors in stored or transmitted data Reconstructing flawed or missing data sectors Allowing more simultaneous read operations and higher performance than Data Guarding (Raid 4) For example If a drive fails, the controller uses the parity data and the data on the remaining drives to reconstruct data from the failed drive. this allows the system to continue operating, although with a slightly reduced performance, until you replace the failed drive. Distributed Data Guarding requires a logical drive with a minimum of 3 physical drives. Therefore, in a logical drive containing 3 physical drives, distributed data guarding uses 33 % of the total logical drives storage capacity for fault tolerance, while a 14 drive configuration uses only 7 percent. Tony Platt

Tony C wrote in message <931333286.482062@cacheraq001>... >Not even close :) > >Raid 5 is NOT mirroring. Raid 5 does an XOR of the data spread across >multiple drives and stores that result on the parity drive. The parity is >usually calculated at the block level and the data is stripped across the >drives in the Raid set. When a data drive fails, the data on that drive or >volume can be 'regenerated' by XORing the parity data with the data on the >surviving devices. This is a description of RAID-3. RAID-5 is basically identical, but there is no 'parity drive', instead the parity location, rotates through the members of the drive set. This makes performance better in random I/O, but worse in fast consecutive block fetches. ><rest snipped, as 'spot on'>

OK time for some ACCURATE information! I know Obnoxio has been waiting for me to join in. Tony Platt wrote: > > Tony C wrote in message <931333286.482062@cacheraq001>... > >Not even close :) > > well it helps if you give the RIGHT advice > > >Raid 5 is NOT mirroring. Raid 5 does an XOR of the data spread across > >multiple drives and stores that result on the parity drive. The parity is > >usually calculated at the block level and the data is stripped across the > >drives in the Raid set. When a data drive fails, the data on that drive or > >volume can be 'regenerated' by XORing the parity data with the data on the > >surviving devices. TRUTH!!!!!!! Tony's comment about the parity striping is just WRONG. The parity is not striped, unless we are just using different terms for the same thing. Here is the straight poop. RAID5 is the same as RAID4 EXCEPT that instead of a dedicated parity drive containing all of the parity blocks the parity block is rotated round robin over all of the drives in the stripe-set. So block 1's parity may be on drive 5 (in a 5 drive RAID5 set) then block 2's parity is on drive 1, block 3's parity on drive 2 etc. round and round. RAID 5 always calculates, for a particular logical block, which drive contains the data and which drive has the parity and reads ONLY the corresponding block from those two drives. Therefore the other N-2 drives in the RAID set are available for similar 2-at-a-time reads and so RAID 5 performs better on small random reads than it's cousin RAID 4 which does better at sequential read since it always reads all N drives every time. Most RAID5 systems use a large RAID block size to improve sequential read performance (64K+). Unlike RAID3 and RAID4, RAID5 does not EVER check or verify parity for data reads. That means that if a drive becomes flaky (that's a technical term) and the media begins to return garbage (yes it can even happen on modern SCSI drives with automatic sector remapping, lost remap lists are finite you know!) then not only will the RAID5 set return garbage but when you write back the parity will be recalculated with the garbage an become trash itself. > Wrong, you are thinking of DATA guarding (RAID 4) > Raid 5 Distributed Data Guarding stripes the parity data across ALL disks in > the set (like the name says) > > Raid 5 offers > Best protection RAID 5 offers NO PROTECTION against multiple drive failure or against partial media failure. ONLY RAID1 and it's derivative RAID10 can be called BEST PROTECTION. If a RAID1 drive becomes flaky the mirror, which is written independently, will be fine and can be used to build a replacement for the flaky drive. > Increased throughput Over what? Over singleton drives? Yes. Over RAID0? No. Over RAID1? Yes. Over RAID10? No! No! NO! > Most cost effective Actually exactly as cost effective as RAID3 and RAID4 which are better for databases than RAID5. Many RAID systems now calculate RAID3&4 parity in hardware on the controller so the main objection to RAID3&4 is no longer valid. And RAID3&4 do NOT suffer from reduced write performance as RAID5 does since all of the data blocks in a stripe block are read together (drives are spindle locked) there is not additional read needed to calculate parity before writing. Since parity and all data blocks are read concurrently for ALL reads RAID3&4 systems can, and most do, check parity on read which can trap and correct most partial media failure problems at read time which is when they are still correctable. > Striping parity data generates redundant information that can be used for > the following > > Detecting errors in stored or transmitted data NO NO NO NO NO NO NO NO NO NO NO NO NO NO NO NO NO NO NO NO NO NO NO NO The RAID5 standard does not require, nor does anyone implement, parity checking on reads to detect errors in stored data before transmittal to the application. Some RAID5 vendors have tried to work around this very real problem by including, as DG Clariion does, block level parity on every data block (DG includes 8 bytes of error correction code on every disk block storing 520 bytes rather than 512 but the Clariion firmware eats the parity bytes before returning data to the host) with differing levels of success. Even with this I have experienced data loss from partial media failure on such RAID5 systems. > Reconstructing flawed or missing data sectors Nope. RAID5 CANNOT DO THIS. NEVER COULD. NEVER WILL. > Allowing more simultaneous read operations and higher performance than Data > Guarding (Raid 4) TRUTH. (Hey even a broken clock...) ;-) > For example > If a drive fails, the controller uses the parity data and the data on the > remaining drives to reconstruct data from the failed drive. this allows the > system to continue operating, although with a slightly reduced performance, > until you replace the failed drive. TRUTH. > Distributed Data Guarding requires a logical drive with a minimum of 3 > physical drives. Therefore, in a logical drive containing 3 physical drives, > distributed data guarding uses 33 % of the total logical drives storage > capacity for fault tolerance, while a 14 drive configuration uses only 7 > percent. More TRUTH. Look for absolute best protection use RAID10, also known as RAID1+0 which is different from RAID01 or RAID0+1, which also gives the absolute best performance and yes costs the absolute most $$$$$$. That is truth. With the price of drives today there is little excuse to say: "I cannot afford RAID10 I'll user RAID5". If you absolutely cannot afford the extra drives for RAID10 then use RAID3 or RAID4 which provide improved write performance (up to 90% better) over RAID5 and do not suffer from partial media failure problems. You are still at risk for losing more than one drive in a RAID set before rebuilding the first down drive but if you are careful to use drives from different manufacturer's lots you can reduce the risk of that somewhat. Art S. Kagel PS: Happy now Obnoxio? I just thought I'd stay out until the level of disinformation had hit a high and then come in as the voice of reason. That way I don't have to shout as loudly. :-)

Finally someone got it right.... :-) All you wrote it true. I just want to emphasize one point: I have help to design and implement a lot of systems which used RAID, in many cases the customers preferred the RAID-1 over RAID-5 solution because they could implement it by SOFTWARE and reduce the cost. This may have some side effects: A. The writes will be slower, since we actually write twice. B. Increase CPU load. C. Might reduce I/O throughput. but, if one uses a HARDWARE controller, we can avoid the above. -- Eyal Lupu Senior UNIX and OpenVMS systems administrator In article <931333286.482062@cacheraq001>, "Tony C" <waclark@noneed.aa.net> wrote: > Not even close :) > > Raid 5 is NOT mirroring. Raid 5 does an XOR of the data spread across > multiple drives and stores that result on the parity drive. The parity is > usually calculated at the block level and the data is stripped across the > drives in the Raid set. When a data drive fails, the data on that drive or > volume can be 'regenerated' by XORing the parity data with the data on the > surviving devices. > > As to the original posters comment that Raid 5 will be faster, this is not > necessarily true and in most cases Raid-5 will be slower than Raid-1 > (mirroring). This is especially true in when there is a lot of write > activity. The reason is that when data is written to a Raid-5 volume, the > parity for that Raid rank must also be read, recalculated and re-written. > This may also involve reading the other data volumes in order to do the XOR > calculation. Mirroring on the other hand only requires 2 writes and if you > use a caching Raid controller this can be very speedy since the writes go > into cache and are later destaged to disk when the disk is less busy. > > There are other factors to consider as well such as adding drives, > increasing the size of the Raid group and so on. Raid-5 is usually much more > difficult to add-on to. You usually have to back-up your data, rebuild the > raid-5 group and then relay the data back onto disk, which btw usually > requires a full rebuild of the Raid-5 parity information. > > Good luck > Cheers > TC > > Salem Lee Ganzhorn wrote in message <7lu7t5$q3l$2@uni00nw.unity.ncsu.edu>... > >sandeep.singh@usa.net wrote: > >: Thanks y'all for your comments/suggestions. I guess > >: I must be missing something cause I did not get what > >: a lot of people talked about using RAID5. RAID5 will > >: no doubt be faster but then there is no redundancy. > >: Maybe two RAID5 channels and then database mirroring > >: across the two controllers? > >: > >: Will someone please comment on this: Suppose I have > >: a choice of having one single RAID1/0 (or even just > >: RAID1) controller OR having two Ultra2 SCSI channels > >: and then use Informix mirroring across the two cont- > >: rollers. Which one makes more sense? > > > >Raid 5 is mirrored. There is enough data mirrored on the disks so if any > >one drive goes down you can rebuild the drive completely from the remaining > >drives. You get the speed of striping plus the fault tolerance of > mirroring. > > > >Of course you only get 1/2 of the space the drives have to offer. > > > >The only downside is you have to have at least 3 drives. > >Regards, > >Salem > > > > Sent via Deja.com http://www.deja.com/ Share what you know. Learn what you don't.

Eyal Lupu wrote in message <7m35si$o4u$1@nnrp1.deja.com>... >Finally someone got it right.... :-) > >All you wrote it true. I just want to emphasize one point: >I have help to design and implement a lot of systems which used RAID, >in many cases the customers preferred the RAID-1 over RAID-5 solution >because they could implement it by SOFTWARE and reduce the cost. This >may have some side effects: >A. The writes will be slower, since we actually write twice. Only for software/firmware RAID and not hardware RAID. >B. Increase CPU load. Only for software/firmware RAID and not hardware RAID. >C. Might reduce I/O throughput. > >but, if one uses a HARDWARE controller, we can avoid the above.

the more separate controlers you have the better the performance. I wouldn't go to a RAID unless you want to have a really gradular control as to what data gets written/access from where. But remember you can't only do this with RAID level 5 -- it handles the disk striping for you. sandeep.singh@usa.net wrote: > Hello all, > > I am trying to set up an Informix database (7.3) on > Linux (Redhat 6.0) on an Intel box (dual PII 400). > Since this is going to be a production machine, I > will be using mirroring of database chunks. I want > to get some advice from people who have setup or > have considered setting up such a configuration as > regards to disk storage: Will I be better off using > RAID or should I use dual SCSI controllers and just > mirror the database chunks. Does it make sense to > mirror the database chunks on a RAID? > > Lets assume that cost is a not a factor in the decision > -I just want to get the best possible setup. I have been > considering the Mylex 960 DAC for the RAID and Adaptec > 2940U2W for the SCSI controllers. Thanks in advance for > your time. > > Sandeep > > Sent via Deja.com http://www.deja.com/ > Share what you know. Learn what you don't.