BAARF

BAARF
Enough is enough.
You can either join BAARF. Or not.

BAARF
# Name Comment
0+1Juan LoaizaWrote the SAME paper, and has argued for many years against RAID-5.
1Cary MillsapWrote articles against RAID-5 many years ago.
2Jared StillHas a "No RAID-5" hat which he wears on a regular basis
3James MorleAuthor of Sane SAN
4Tim GormanFounder of People's Front for the BAARF - radical, militant splinter group
Raid 5Peter GramHas helped many customers out of their RAID-5 misery over the years
6Mogens Nørgaard
7Tim Onions
8Alex Høffner
9Gunnar Bjarnason
1+0Torben HolmGroupie
11Carel-Jan EngelsFounder of RFLF (Raid 5 Liberation Front) and PMSS (Poor Man's Storage Solution).
12Morten EganConvinced the Oracle Raw Iron Development team to abandon RAID-5 and use RAID-1+0 in stead.
13Leif Barbré KnudsenSpecifically asked to become member no 13 before member no 12 was found.
14Kim Poulsen
15David Kurtz
16Richard Ji
17Stephen Andert"I've alienated so many SA's that they could start their own country <grin>"
18John KanagarajHave been fighting RAID-F for about 3 years now.
19Anjo KolkHas placed the BAARF logo on OraPerf.com
20Barbare BakerDo NOT believe the liars and infidels. We have no BAARF in our computer room, nor will we allow an invasion of BAARF.
21Rachel CarmichaelI've been fighting RAID-F for what seems like forever. Not usually winning, but fighting. Mine not to reason why, mine but to fight and die
22Frank Pettinato
23Brian PeaslandI've been fighting Raid-F for many, many years now. I currently have a multi-terabyte system which we want redudancy, but the powers-that-be won't fork over the cash for RAID 1 ro RAID 0+1. So RAID 5 it is. I told them that data loads would take twice as long. And sure enough, they do. But they still insist on RAID 5 as the RAID level of choice. Ugh........
24Paul BaumgartelWhen I arrived at my new job, I found RAID 5 everywhere, and a sys admin who wanted to build my new database servers that way...I smote the old systems and set the new ones on the right path.I would be proud to be associated with your movement.
25Bill "Shrek" ThaterOK i'm in. I've been at this for a while, but damagement would rather believe the salesdroids that they guy actually having to deal with the stuff.
26Dick GouletOur storage team won't even respond to me anymore when I ask for the manufacturer's rating for non-cached I/Os per second & number of controllers, RAID level, striping, etc... All I get is 'why do you want to know that' and 'what application is this for'... BARRF will stop the debilitating headaches since I will just refuse to talk about it anymore. Our storage team won't even respond to me anymore when I ask for the manufacturer's rating for non-cached I/Os per second & number of controllers, RAID level, striping, etc... All I get is 'why do you want to know that' and 'what application is this for'... BARRF will stop the debilitating headaches since I will just refuse to talk about it anymore.
27Don BurlesonI have never seen any reason not to use RAID 0+1 (RAID10) with Oracle databases, but plenty of shops suffering because of ignorant vendor recommendations.
28Peter McLartyI was introduced to database servers in 1997 starting with a SCO server with Progress that was running RAID 5, For the size of the company it didn't run too bad, but as I learnt to manage that database. And learnt the benefits of having RAID 0+1 I decided that when I upgraded the system I was having no more RAID 5. My boss gave me the budget and I got the new array and lost the RAID 5. Performance went up splendidly.
29Gudmundur Bjarni Josepsson
30David Phillips
31Geoff Smith
32Nils BøjdenFinally somebody speeks up for us poor sods trying to get the databases to perform decently. Use the RAID F# for word document, speadsheets or pictures of women dressed interestingly, but never ever for the databases.
33Michael BrownOkay, I give up. As much as I try to stay out of these discussions, I just got drawn into one on OAUGNet-DBA. Give me my member number.
34Leif ArildsenI think it should be called RAARF (Raid Against Any Raid Five, Four, Free, Fone.
35Yannick GagnonI'm fighting against those DBA of the outsourcing side that are putting all production system(mostly OLTP) on big RAID-5 on SAN (those big Hitachi one) justifying it that they don't have any perf problem since they can put up to 32 GB of write-cache($$$).
36Daniel FinkI think that this should be RAID-5Z (Z is the closet character that is a 'mirror' image of 5).
37Bjørn EngsigUndskyld! for having failed in the Battle
38Art S. KagelI've been fighting the anti-RAID5 fight since 1995 and have just one thing to say: NO RAID5 NO RAID5 NO RAID5 NO RAID5 NO RAID5 NO RAID5 NO RAID5
39Bjarne Hakonarson
40Tue MarkerHas suggested a course in how to argument against your storage vendor.
41Paul ValleeI've fought fo many battlef against RF I've loft count. I've loft fome of courfe but won my fair fhare, and I've got fe fcarf to prove it!
42Ron Thomas
43Piet de Visser(Raid, the Cache, and everything)++
44Tim JohnstonHey... I'm jealous!!! BAARF! BAARF! BAARF!
45Jesper Haure NørrevangI want to be a BAARF member now - I had a meeting with a RAID-5 salesman yesterday.
46Don GranamanStill waiting for RAID-42…
47Robyn Anderson SandsI'm a DBA with many battle scars from fighting RAID-F and other bad implementation choices. I've been called a primadona and a technical purist and taken lots of abuse on this subject. I'd love to put your logo next to my UNIX Live Free or Die license plate ...
48Kimberly L. TrippWow - I like you Oracle people more and more the more I know you! It's really only recently that I've taken on Oracle people as friends and so far I like most of you (except for those of you who are bad influences in terms of beer :)...
Seriously though - it's not surprising that all of us think similar thoughts for databases and unanimously agree that RAID 5 is poor man's fault tolerance and a great sacrifice of cost over performance. In the end it's just not worth it!
Ah, I feel better already! I guess this is the first step......
49Ole WissingOnce upon a time in a little country, not too far away, there was a RAID 5
which in the beginning was based upon 12-14 drives including 1 hot spare. Everything went fine.
When a drive crashed, the spare was eager to help and took over the work work without any problem.
The IO performance was not an issue, because this system was only feeding a fileserver.
During the time the array was expanded and expanded and ended upon 25-30 drives that all lived together in total harmony.
Then, one hot summer day in July, two drives crashed at the same time and the RAID was totally crashed because the risc 0/00 was the same as before, but now it should cover many more drives.
The array had to be rebuild from tape = 800 Gb = 1 day downtime = expensive. Did anybody learn anything, Nope, RAID 5 is still being used in arrays with an ever increasing amount of drives.
When will they ever learn ?
50Carl BruhnI'm in one of those outsourcing companies, but we can think, read and understand
(at least some of us :-) So for me and my databases. No more RAIDF. ( and even better end of discussion).
51Lon WhiteYou're going to be using RAID F... Why not, after all,
what would an Oracle DBA know about the best setup for Oracle Databases???
52Rick GoudeauI have to agree with BAARF principles. I have spent much time arguing with with management who have been sold snake oil about raid 5.
Since impact of Raid 5 is so small, it is curious why rebuilding the file systems to raid 0+1 significantly outperforms the previous environment.
Oh well on to the next battle!
I think the BAARF party makes more sense than the 2 major political parties in the USA.
53James PettsA sysadmin who forces RAID-F on a poor suffering DBA is no better than a car-hire company that will only rent you an Edsel
54Hans ForbrichRAID - one of the few cases in modern math where 0+1 is truly greater than 5.
Just tell the accountants that they can do their own backups and restores if they think otherwise.
55Andrew HammIf a supplier screws up the configuation I just tear down and completely reinstall, and then tell the engineers what I've done.
They probably go pale and faint over their wasted hours, but they never do it again.
Actions speak louder than words!
56Eckard PotgieterAs you all know, afraid stands for -> against five-raid, so all I can say is: Be afraid, be very afraid.
57Jeff EberhardFinally had a chance to use BAARF. Sys Admin wanted to know if the new disks should be configured as Raid 5. BAARF!!!!!
58Henrik VerupI would like to join the Baarf world, so I can learn how to advise customers not to use Raid5,
avoiding getting into too many undskyld-sessions afterwards.
59Hans DriessenAgain helping out another customer who needed perfomance, but was forced to buy Gbytes instead...
60Steve MontgomerieWhere's the link on how to dispose of overzealous raid-5 UNIX admins?
61Thomas DayTired of explaining to customers that RAID5 does not mean that you don't need backups.
62Paul DrakeThe last storage subsystem I specified had 42 drives mounted on 3 dual channel SCSI-RAID controllers, a mix of RAID 10 and RAID 1. RAID 5? Not on my watch.
63Mark W. FarnhamClearly, I have long since tasted more than enough vomit in my mouth due to useless blather regarding RAID-F***.
Long live Baarf in its quest for less barf.
64Prashant Khanolkar
65Radu CauleaI have to deal almost every day with RAID-5 black boxes (read optimize Oracle databases already installed on this type of architecture) and I am sort of tired of customers' surprise when hearing 'Oracle mirroring of redologs means more than raid mirroring of redologs'.
Long life to BAARF.
66Christophe TsobgnyI have been working on several implementations and it seems like Raid 5 has always been mandatory, the funny thing is that many years ago after reading about different types of Raid I was wondering why we did not implement Raid 0+1 (Raid10). It was really funny because i was convinced at the time that stripe and mirrorIt was really funny because i was convinced at the time that stripe and mirror was the best choice but somehow top level decisions always made us implement Raid 5 - it was as if we would received the storage systems with Raid 5 already installed.
67Diego PafumiI hope these docs will help all DBA's to undertsand how CRITICAL is to avoid RAID5.
Most of DBA's and Sys Admins has NO IDEA about RAID5 performance.
68Thomas HedemannThat’s it!
I’ve had it.
I’m in.
From now on I will simply tell the sysadmin that “I refuse to have a battle of wit with an unarmed opponent!” and not exchange another word with him until he’s read the BAARF articles.
All of them!
69Joe TestaI figure I have to wait for a while to get BAARF party membership #69, bwahahahahahha
70Michael Brown#70 Very good. It's the IQ of my sysadmin. You know him. He's the one that's constantly touting Raid-5 to management.
71Maheswara Rao
72Lars Bo VantingI thought this RAID-5 discussion had been ended long time ago and storage vendors had become serious, but I was wrong...
73Ashok Sharma
74V Rao Kanneganti
75J Muller
76Rich HollandSign me up! I hate RAID-5 with a passion.
77Piers Truman-Baker
78Peter LynchAs a freelance performance specialist who earns a living fixing slow databases, I say RAID-5 is great. Let's have more of it everywhere and keep me in a job!
79Phil Kurjan
80Dennis Heisler
81Raj Jamadagni
82Sten RognesDespite my Viking-blood, temper and physics I've lost my battle against RAID50 here on the other side of the pond. I am with you.
Can I please join the BAARF party?
83Larry WolfsonSign me up!
84Senthil Anand.G"I'm Senthil, a victim of RAID 5..
I'm working as an oracle dba in National Investment Funds Company, Muscat...
its a monarch's company with tons of money and no brains...
I joined up at the fag end of a new project implementation....
and the machine is on a IBM cluster with a raid 5 as the central storage...
and unfortunately for my horror this is a RAID 5 on a pool of 3 disks...
now this is a real disaster for me....
all the files including logs, control files have been dumped on the same array....
and moreover the database is to support a hybrid application ....
and the decision makers out here for the hardware feel it as unprofessional to comment on their decision of hardware...
This is real hell....
85Mark Bobak"If there's enough cache it always works fine!"
86Konstantinos Hairopoulos...a part of my job is system integration with Oracle and Unix platforms, another part of my job is trying things to run faster.
RAID-5 a favorite myth for scalability, reduntancy. I am very glad joining this group....."
87Jan Overgaard
88Rune RasmussenI need to become a member - after 20 years of databases on raid 5 with the never ending "what can we do to improve performance"
I'm ready to join anything that will combat it.
89Cord LangeWe want to join your BAARF world, because we made also bad experiences with RAID5 systems.
We think the worst idea is a ATA-2-SCSI RAID 5 system on a single channel...
We use SCSI or FC drives as RAID10 with min. two channels plus hotfix drives - never made bad experiences with that ;-)
90Bernhard GrambergI am ready to join, because RAID-5 has two other drawbacks:
1. people forget backups.
2. nobody cares.
Re 1.) One answer to the question of a customer, having lost all data, was: "RAID-5 is my backup!".
Re 2.) Nobody cares: And another two urgent questions: Who knows, that the first drive of Raid-5 crashed ?
Nobody. Who knows, that the second drive of Raid-5 crashed ?
Everybody!
91Sam GentschSign me up,
I fight this battle daily at work...
Everyone is convinced the cache makes up for any RAID5 penalty and that RAID 5 is so much safer than 0+1...driving me nuts!
92Mark StricklandI want to understand clearly the issues with Raid-F so I can communicate clearly to decision makers in the client companies in which I work.
93Duc NguyenGood. Simple
94Sassan KaraiTerabiting Raid 5.
Grey hair says there is no cheap, large, fast disk space.
Dont starve your DB server with slow disks, feed its memory.
Make it BAARF = better architecture and real fast
95Morten AmstrupNow, I'm not an actor (DBA = Death Becomes the administrating Actor???), but I would still claim:
Performance is everything!
Let's BAARF!!!
96Christopher GaitI'm joining BAARF because Mogens said so.
Danes used to be great hulking Viking types who used axes as toothpicks, so you simply don't want to say no. Oh, and also I've attempted to fix up one too many production instances to find that their redo is on RAID5. BAARF!
97Andrew MobbsI never again want to have to explain why RAID 1+0 is faster and more reliable for any given number of spindles for this workload.
It's easier just to tell the implementation people a kitten dies every time RAID-F is used.
98Henrik RasmussenHeard about the Parity/Pain stuff about a year ago....
Oh, and your database who has many writes every day and runs on RAID F doesn't perform properly - no shit, Sherlock?
99Mario BroodbakkerIf there's enough cache it always works fine!, he said while looking at log file write times of 50ms and more
100Mark MorrisAs chief architect for storage for Teradata,
I pretty much singlehandedly killed off RAID-5 in favor of RAID-1 for Teradata in 1999 as the transition from 4GB to 9GB disks occurred.
Yeah, there were a few die-hards, but anyone still using it with today's disk sizes is insane.
I believe the case can be made that Patterson, Gibson, and Katz set the storage industry behind by a decade with the lousy ideas of RAID-Fee/Fo/Five.
101Gaja Krishna VaidyanathaAs the guy who played the "lead role" in the BAARF Musical,
having read the weirdest script in my amateur acting career and having sung from freaky modified lyrics of popular QUEEN songs,
I don't see WHY NOT!
102Connor McDonaldBecause 102 is 42 in base 25. And because SAN's are driving me inSANe
103Kevin FriesThis stuff just kills a system. BAARF on.
104Colin BullI have just been given this gem by a salesman - RAID-DP (double parity) basically gives 4,500 times the resilience of RAID-4. yes,
I believe it.
105Bob McAuslandClient : I want resilience, what do I need? Money no object.
Me : Disks are cheap now, go for mirroring, not R5.
Client : So R5 is cheaper?
Me : Only marginally, but mirroring gives you .... Client : So its R5 then!
106Ally McIntyreCurious. It has the same sound and subsequent nausea that I get when I have drunk too much.
107zach friesebaarf!
108Claus Busk AndersenEnough IS enough!
109Aaron NelsonBAARF!
110Van Driessche ErikWhen will R5 be dead ?
111David McCarronI accepted RAID-F and went with the flow. All I had to do was forget anything I thought I knew about disks.
This all came about when one of the engineers from our prefered hardware supplier told me it was not possible to create a partition across the outer edge of the disks platters.
I gathered from this statement that disk technology had evolved so much that it now lived up to the glossy brochures and the sales pitch.
What with 2Gb of cache and expensive fibre disks, what could I possibly have to worry about
( Q. Is the fibre used in the bearings of the disk spindles to make them spin faster and with less friction, so reducing latent seek times ?
That must be a question for "The Challange", maybe I'll win a T-shirt !! )
But it has proved to be more efficient in one sense, now I don't have to do so much analysis to find out what my database is waiting for - I always know where to look first.
Next I'm going to get a Solid State Disk array and put RAID-F on that too ! I love it !!
112William ShipwayF time has been and gone; let's get back to the three R's - Realistic RAID Recommendations.
113Lee McCannWe have no choice but to place our UNDO & TEMP & REDO on RAID5 because that is all we have; BARRF!
114Simon ColeAhhh - I love the sound of a crammed cache in the morning. Sounds like victory...
115Bassam ZahranInterested in RAID
116Peter James HitchmanLost the argument long ago. The budget rules OK.
117Yuri van BurenThe Battle has to continue!
With disksizes growing and growing we get much more data with much lesser throughput.
IO will be the main bottleneck for scalability the coming years.
118Stephen BoothWhy, oh why, do a certain storage vendor keep peddling their RAID4 arrays? And why do managers listen to them.
119Robert de LaatNever too late to join a club with common sense
120Yoann MainguyTired of "tunig" databases over Raid-F. Tired of sysadmins who don't really try to understand.
121Todd BourneI have been battling SAN Vendors, Managers, Disk Admins, Sys Admins and DBAs for years. RAIDF is BAARFING madness.
I am sick of vendors and admins saying "trust me.. striping and caching will fix everything".
Enough is enough!
122Benny Arbjerg PedersenNo more 3-4-5
123Mike BadarJust say no to Mr. Man. Those who know, know. Those who don't follow the cattle.
124Sven SchuranI have ask someone, what is best for my database, he said RAID10.
I checked it in the Internet and I got the Opinion it is true. Later we had to buy new Servers, a oOnsultant said, just use RAID5.
I said neverever and got my Raid10. And know i found this page and have to join.
125JP NandiGood One for KM
126Les HollisBeen against RAID5 since I started as a DBA in '92. "Give me 1+0....or give me death" (to paraphrase a great american hero)
Current position we are using RAID5 but my incessant bitching finally won them over. We have a new 20TB array on the floor that will be striped 1+0.... "VICTORY to the masses!!!!!!"
127Joel GarryIt's about time for RAID-5 to be more generally accepted, so must be about time for me to be a dinosaur against it.
128Anthony McCollumI currently have my first Raid5 system.
Performance is not good.
I work in the spatial arena so writes are very large and the penalty is very heavy in the R-5 config.
Im looking into R-0+1 config as we speak.
129Michael J. Hillanbrand III hate RAID almost as much as raw files - but given the choice I'll take raw on LOTS of disk and manage the reliability and replication myself.
130Juan de VillerosI have suffered raid5 data failure twice in my company. Every time I was able to say "you see! raid5 is not secure". Now everything we do is mirrored.
131Jonathan Van HoutteHave been agitating against RAID 5 ("the greatest thing since striped bread") for years.
132Ole ThomsenBetter late than never
133Ian CarneyHallas accolyte and founder of the Provisional Wing of BAARF. I also went university in Baarf...
Once had a argument with a Sequent engineer who thought I was mad - he was probably right but not in this context
134Jim AveryI've had enough of poor performance, especially when doing anything which is bigger than the huge cache on our system.
135Dave ScottHaving seen a great presentation at Oracle UKOUG 4 years ago involving glasses of water to demonstrate raid levels -
I have been a firm convert to the anti RAID 5 et al brigade
136Bernard (DUL) van DuijnenThere are already enough databases to repair, I do not need raidF for more work.
137Teresa SchoenCan't pass up joining any organization with such a creative name :-)
138Robert FenstermacherEnough IS enough
139John A. KostelacI see client after client who ask me to help them make their systems perform better and then reassure me that I need not look at the IO subsystem as it is RAID5. Invariably, that is where I find signifcant problems. Oh well....
140Frank KalisThe more I get to know about RAID, the less I see a use in RAID-F.
141Augusto Cesar Vianna de CarvalhoNo RAID! No RAID! No RAID! No RAID!
142Chris Du-PondSimply put, I had enough issues with customers in RAID-5 that I have come to hate it...Plain and simply it is not a good solution for Oracle despite large caches in newer storage array systems..
143Suraj KrishnanHave recommended against Raid 5 to atleast a dozen customers.
144Geoffry TrembleyTired of Fighting NetApp Salesmen
145wan chiun keatGreat info about the RAID. Recommend by someone from Forums
146Vijay CherukuriWorst experience with RAID5
147Arie MarsMostly because of the beer or two...
148Gary Alan MyersThis is GREAT STUFF!!!
I have argued against RAID5 for years and years and referred people to Oracle's "SAME" protocol (Stripe and Mirror EVERYTHING) ...
but some customers are still sold on the idea that a "30%" disk integrity overhead is better than "100%" ...
until, of course, they have a catastrophic drive failure; and why does it always seem to be the parity drive that goes out!!!???
149Brian KushWho cares about RAID anyway.... I only have one drive in my laptop.
150Crispin ProctorMore needs to be done to let people know that the "default" for arrays should not be RAID5. I have had 2 Catastrophic failures with 5 (200GB volume) and NON with RAID1+0 on a 12TB volume!!! Long live 1+0
151Richard JacobsBe different SAME is good!!! Long live ASM!!! A seasoned campaigner against RAID 5
152Martin GamtofteJust cause.....
153Erling SkaaleIt's better to be fast than wrong
154Steen Vincentz JensenUsed to believe that noone would ever dream of using a worse kind of Raid than Raid 5 - but have now discovered that one of our customers is using a SAN with Raid 53 - and that is worse!!!!
155Ronald GillMy main reason for joining BAARF is that I stayed in the Grange hotel where it was founded just about a month before the founding, oh yes,and I really hate RAID5.
156Nelio KuboInformation is Priceless. Performance and High Availability are everything.
157Antoni QuintarelliStill being confronted with organisation using SAN RAID5 technolgy and 100% disk utilisation.
Will they ever learn.....
158Leandro Guimarães Faria Corcete DUTRASometimes it just take guts to take a stand against nonsense... now if only someone would do the same to all this 'SQL is relational' and objects nonsense
159Rasmus JessopComments : I just happened to come across this "nice" little note on Metalink
Note:38281.1 What types of files are suitable for placement on RAID-5 devices?

Placement of data files on RAID-5 devices is likely to give the best performance benefits, as these are usually accessed randomly. More benefits will be seen in situations where reads predominate over writes.
THIS JUST HAS TO STOP !!!
NO MORE RAID-5 HORROR STORIES !
160Ted CoyleI've seen RAID5 up close, and I don't like it...
161Zoran GagiKeep the storage vendors honest :) - with a large random IO working set SAN cache is close to useless - buy lots of spindles and use RAID10
162Petri TumppilaPlease teach the storagefolks to use calculators. They'd earn more if they'd sell raid10 instead of 5.
163Taral DesaiI want to be a expert in this oracle world of data
164Michael DinhI just so tired of brain-washed system administrators who don’t know anything other than RAID5.
165James L. KendallI have believed in this since I read Cary's paper back when it was an internal Oracle document (I WAS an employee at the time!).
In the 10 years since, I have learned to just allow people to make the mistakes that they are determined to make - sort of as if they were teenagers!
166Brian R. DayDevelopment support for SAP Inc. managing all Oracle customers. Subcontracted to SAP from Oracle Corp. RAID 5 .... ARGH!
167Douglas E. MarshStick it to the RAID 5 man brother!
168Neil ChandlerThe SAN administrator will only ever ask my how much SPACE I need.
169Jeroen LangeveldHeard enough nonsense from so-called SAN experts...
170Brunell MartineauRAID five is old and it sucks.
171Tony RogersonYour site is excellant and just what the RDBMS space needs, i'm sick and tired of people going the RAID 5 route, all too many 'sys admins' think file server - arrrrr, you also need to set a site up explaining what cache is too and why a program, in my case SQL Server takes so much memory - often called 'a memory leak' by sys admins.
172Werner KirschI was totally pleased on a whole community sharing my personal problems with RAID5 and the vendors behavior of neglecting all complaints with statements like 'the cache is big enough to handle your requests' or 'we have no problem with your random writes'.
173Vedran DinterGo BAARF!
174Casey JordanI am a Oracle DBA that has been fighting RAID5 for years. Just when I thought like I coudn't fight anymore I found your site.
I am now imspired to keep on fighting.
175Mark J. KounalisIBM - Shark - do I really need to say more?
176Li-Shan ChengSAN Admin: Ok we have RAID-5 or RAID-S, they are similar. Currently we have 6 TB in our EMC DMX2000 SAN, how much space do you need?
Me: Please can I have my space configured in RAID 1+0?
SAN Admin: No, we waste too much space that way.
Me: But I only need 400GB for my transactional system!
SAN Admin: Talk with your manager, these disks are very expensive!
177Örjan LundbergAfter spending one week to prove that RAID10 is what the customer wants for their Streams AQ env (10GB or so of queue data) it is good to find a whole site dedicated to the data.
(This is people that wanted to disable the array cache since they where afraid of loosing data in the cache)
178Keith MassnerNO NO NO!!! No Raid 5!
179Mike WojcinskiI've spent way to many hours arguing with Sys admins and DBA's discussing the merits of RAID-10 and the cons of RAID-5... DIE RAID-5 DIE... DIE...
180Klaus DittrichI'm just fed up discussing "What's cheap now" instead of "What's best in the long term"...
181Joze SenegacnikMy past experiences with RAID5 used by my customers forced me to join BAARF.
182Sean IvusicHaving information to convincingly convince clients to use RAID10 / 1+0 / 0+1 rather than RAID-F (5,4,3) helps significantly. Besides, I hate RAID-5 ;)
183Jan-marten SpitRAID5 should be regarded as a weapon of mass obstruction. Any RAID5 system can be preemptivly invaded and burned down to the ground.
184Frank FekkesMost of the SAN now goes Fm it is becoming a disease.
185David BuchholzNO RAID 5 !! Databases love RAID 0+1. May RAID 5 return to the dark abyss from which it was first conjured up, never to be seen or heard from again.
186Guido Jimenez M.Nice web page.
187Jim HambletonIf raid-5 did not exist, we would not ever have invented it.
188David LeeExcellent sentiments I could agree more, RAID-F is rubbish
189John Fjelsted LarsenThis war is not over yet, BAARF is needed now more than ever.
190Frank HamersleyAt last my decade of paid subscriptions is recognised!
191Brainiac 5Folks, you can't imagine how proud I am of finally being able to sport a RAID 5! I dreamed of that for 5 years, at the least. So how can you dare to burst my beloved bubble? And you know what, I don't care how dangerous it is compared to RAID10 - I used to go single-disk-no-backups all my life and this feels so much better now. So consider me the official mole. I will ignore all warnings, not back up, rely on my RAID5 and promise to post my all-data-lost horror story soon. Well. Or maybe not. Thanks for the education. And the burp, eh, baarf. I joined because of the cool name I guess.
192Frits HooglandThe RAID-F keeps popping up everywhere. With SAN/NAS, we enter a whole new era of IO tuning.
193Albert BertilssonBecause it works.
194Phil WestThe SAN people are easily startled, but they'll soon be back, and in greater numbers.
Spent 18 months on one site battling to get access to the physical SAN layout details and fighting the 'It is SAN. It is fast' mantra - and won
195Douglas CoanBasic due diligence into determining if technology advances over the last 5 years makes raid 5 a remotely viable option for a 10TB data warehouse lead me to BAARF in more ways than one. I guess that the laws of mathematics and physics that used to 'slightly hinder' raid 5 still exist.
Perhaps someone will get this resolved once that whole speed of light nut is cracked.
196Thomas PresslieI'm proud to say that I've already won the battle against RAID F at my place of employment - I'm Scottish after all.
I now need to find somewhere else to battle now. Let me know, I'll be there!
197Aaron WermanI am here representing the RIOT-BUST-ROT (RAID Is Only To Be Used Storing To Read-Only Tablespaces) front. We demand parity! I warn you - we will not quietly hand over our caches! Redundancy is the hobgoblin of small imaginations.
198Andrey KriushinSwitching to mirror+stripe helped one of biggest russian mobile phone service provider to meet throughput requirements. Rise by a factor of 4-5. After half a year of argueing. Enough is enough. Never more.
199Erik SwinkelsFinally got rid of the last RAID-f volumes and feel free to join now :-)
200David J. ClarkThanks to RAID-5, I get to be an IT consultant helping companies improve database performance by removing RAID-5 !
201Bob SneedTANSTAAFL, world! I can't say categorically "don't use RAID5", because it's useful in some contexts - but don't complain about its performance. Host-based (versus controller-based) is especially performance-inhibiting! Understand the physics, make informed decisions, and enjoy life!
202Hitesh ThankiBring out your RAIDS
203Yue.GuoI don't like Raid 3,4,5.I only like Raid0+1
204Rob DukeI am tired of fighting the RAID-F battle from a position of having to prove that it is wrong.
205Chad DinermanI am sick of fixing broken raid 5 systems and it's time people learned about the dangers.
206James HamiltonDisk space is cheap. I/O capacity is the furthest thing from cheap. RAID-5 has been killing us for years. Make it stop.
207Kevin ClossonI have at least eleventeen reasons to loathe RAID5. This form of RAID is a wee bit like sliding down a mile long razorblade into a bucket of salt. The only remedy is to drink until it hurts and then BAARF until it feels better. Having said that, I suspect I've been mixing my drinking with my thinking.
OK, I need to count how many successfull systems I've personally been involved with that were RAID 1+0 with hundreds of spindles ... ready, set, go... this is going to take a while ...
208Frank B HansenAfter running a modify script against a Raid 5 DB for 2*42 hours, I have to join forces :o)
209Mike HindsRAID 5 is fine, only if one prefers mediocre performance as well as mediocre redundancy.
At the price of media today, why not the best? More to the point, why not noticably improve our job results?
210Jim CarterAfter losing 3 RAID 5 (aka CRAP - Can't Really Afford Performance) volumes due to multiple disk failures at two different job sites, he fights the good fight each day.
211Jarmo AlataloThey said that Disks for the SAN box is expensive. It's true but think about how expensive it'll be to recover Terabyte database.
212Steve GreshamAfter having to restore a Raid-5 setup by kicking two dead drives with my steelcap boot after the server was in a fire, I denounce all setups where only two drives need to die to make my life hell....
213Marlon DavisTo get assistance whenever vendors convince management to purchase a SAN solution anb use RAID5 only array
214Michael StovekenEnough with ServeRAID, rebuilding arrays, defunct drives, and all the bull that goes allong with striped redundancy. RAID-1 is the only RAID I believe in and after my latest bout with an old town hall IBM Netfinity server and its RAID-5 configuration I will never support another RAID-5 environment.
215Chieu TranIn the mid-1990s I was doing a lot of Oracle benchmarking for a US-based hardware vendor. When management pushed me into using RAID-5 we kept coming last/second last in any benchmark against other vendors. It took me about 6 months to convince people that striping and mirroring was the way to go if management still cared about winning in benchmarking. The sad fact was although we stopped using RAID-5 in benchmarking, the salespeople still flogged RAID-5 to our customers.
216Karl MillerJust lost a prod server to RAID5. All I asked for a 0+1 for 200GB. Hey, world..0+1!=5
217Dave DhirenOh No, No More RAID 5
218Roman KleselEndless discussions, debugging and configuration sessions in order to work around poor I/O due to RAID-5 configurations.
219Jerry HosfordCutting through endless Oracle speak in pursuit of a direct answer.
220Juan Martín GuillénDon´t know if RAID 5 is bad or not, but my dog barks something like BAARF, BAARF!. I'm in.
221Robert L MathewsWe built a (software) RAID 5 array with as many disks as possible to "minimize the cost".
That required a special server that supported more disks than our other servers, of course.
Eventually, the inevitable happened: that server died, and we couldn't easily read the data because we didn't have an unused spare computer that could mount all those disks. With RAID 5, of course, you aren't going to read a single byte unless you can shove (almost) all of the disks into one computer.
If we'd used RAID 1 and lived with the minor inconvenience of multiple partitions (which would work just fine in our case: the server is used for backups of thousands of Web sites, and each site can easily be automatically assigned to partitions with available space), we could have taken any one of the disks and plugged it into any computer and read the data back. Lesson learned; the shiny new replacement server will be using 10 disks as 5 RAID 1 partitions. If the entire server fails, we can easily get the data back by plugging single disks into another computer, and even if three disks failed simultaneously, we'd still only lose at most 20% of the data. Try *that* with RAID 5.
222Ken AdamsonRAID 5 has been the bane of my existence, I want it to die. Stop the insanity!
223Rich JesseTwo words: HP Auto-(bleeping)-RAID
224Flemming DanielsenNo RAID 5
225Ryan BryersI'm a TA, I've worked on the dark side and spec'd RAID 5............
I need to learn something useful now.
226Justin CallisonI've finally found a place where I belong. I'm not alone. There is intelligent life out there. I love you guys ....
227Muralee Anantharaman Easwaran!!!! Trying to be equipped before I get into battle field
228Didier Van HoyeI've been slaying each and every RAID F setup I've come across in my career. Since there are more RAID F's than I can master I decided to get organized and join BAARF.
229Brad SimmonsRAID-F on a SAN? Super-BAARF! Stop the in-SAN-ity!
230Robin HarrisRAID 5 is a kludge. RAID 1+0 is the Rolls Royce of RAID.
231Richard DavidAs an Ingres and Oracle DBA for more years than I care to count, I have grown to loathe the term 'RAID 5'. Urghhhh!
232Pat PriceI too am tired of fighting vendors who influence storage "specialists" who influence DBAs that RAIDf is good for them. I decided to become a card carrying member of the BAARF party as soon as I found out it existed!
233Johnny J. AndersenNo comment, just stay away from Raid F, Okay?
234Gordon MillerI can't have the bad write performance that RAID5 offers in a real-time recording environment. RAID5 should be outlawed under the Geneva Convention.
235Simon HoltDear SAN admins: Repeatedly telling me that the cache solves all contention problems does not make the file wait events I always see disappear...
236Þröstur JónassonJust lost an 6 drive RAID5 array. Do I need to say more?
237Frederik HertzumAfter reading the "Why should I not use RAID 5?" article, it just doesn't make sense to even have the standard anymore.
238Jinran WangI never thought RAID5 would impact the performance that much until we have a RAID5 drive. It is slow slow and slow. As an active Oracle database developer and administrator, I cannot take it any more!
239Martin FarberAwesome! I wish I had known about this site years ago! I've been fighting this battle since RAID went from Inexpensive disks to Expensive subsystems! 8^) Keep up the good work! Thanks!
240Brad HavelRAID-5 sucks. It just plain sucks. Overall it offers a false sense of performance at a lower cost. Well, you get what you pay for.
241Erkki VanninenJust now fighting RAID-5 I/O performance problems on Oracle databases.
242Michael JacobsenI heard this is the site for all answers concerning Oracle, so here I am.
243Jose Figueroa BSimply brilliant! I have spent long time having tried to convince specially to clients with Oracle technology that RAID-5 is the worse election, when have budget to RAID-10 With this page my work will be, much more easy much
244Steven WhiteRDBMS and RAID 5! Need I say more
245Philip PapadopoulosSAME is the only way to go!
246Joel BrungerFed up with dealing with poor performing databases due to poorly configured disk arrays.
247Alan CampbellI am on the BARFF team for life!
248Venkat S. DevrajAuthor of "Oracle 24x7" (Oracle Press) that includes an in-depth discussion on not using parity-based RAID levels for write-heavy applications, regardless of the size of the I/O sub-system cache!
249tapio oikarinen"stupid is as stupid do"
250Bill ReadSo I can impress my boss with all my RAID knowledge
251Krister EvenmyrI'm fed up with HW vendors stating that RAID 5 is now your best option when you want to optimise I/O performance. I'm also fed up with those who have bought it just because the vendor is a well known organisation!
252Shannon StreifelBack in the day, some idiot deployed an Oracle database (redo, data and control files) on a single RAID5 spindle violating Oracle's Golden Rule. Alas, they said they knew better. Until a bad sector in the active redo log file acted up. After I said 'I told you so' and recovered the database, it was redeployed on raid0+1 and raid1 spindles. Eternal frustration for those who use the 'f' word...
253Ernesto Ghettiit's a right battle
254Florin ManailaIf one thing didn't work, I'd just try something else because I knew there was something that would work. There is always something that works. It's just a matter of finding out what.
FFBert Jan MeindersGetting tired of the RAID-5 defenders.... when will they see the light....
256Guenther StuernerEnough is Enough. We are commited to customer success only and not to Raid-Levels which cause more problems.
257Paul JandaTwo Words, first uttered in my "Office Space", in 1998: "Write Penalty"
258Tom ThomsonI have loathed RAID 5 for about 18 years now, have taken my last two employers completely off it (to RAID 10) - why have I only discovered BARFF now?
259Joe MooreRAID 5 never really seemed quite right to me. My subconscious must have been telling me something. I didn't realize other's realized the same.
260sridhar ramachandranGood site with lot of info
261Dave LaPointDon't BAARF and die the death of a thousand i/o's. Live Free of RAID5.
262Pawel PotasinskiGreat site! Great mission! Great jokes ;-)
263abdul fubarRAID5 databases are the single cause of global warming.
264Alex GorbachevRAID-"F" bites sooner or later. It's just the matter of time.
265Bret DunbarI was just reading "Battle Against Any Raid Five" by by Donald K. Burleson, and I saw the mention of your group.
I can't tell you how happy I am to see that there are others out there who feel the same way about RAID 5 as I do. I am a Performance Tuning Specialist, and I have the same argument at almost every site.
I used to work for the Oracle Corp. in their Core Technologies division, as a Senior Consultant back in the 90's. They taught us Best Practices, and that included staying away from RAID 5.
I agree that Cary Milsap's White paper should have dismissed any further discussions on the subject, but I find that the majority of IT managers are not as tech savvy as they were in the 80's and 90's.
Thanks for letting me vent my frustration!
It's GREAT to be part of the team!
267Bryan JonesFor the record, I hate RAID 5.
268David Ballester MontolioSelect randomly a statement of previous members, will fit what I think
269Vesa PajuojaSimply the best !
270Bryan WhiteheadWhy do idiots spend so much money on licenses and fast servers and choke them with raid-5? I want to hit someone.
271Andreas ChatziantoniouSetting: A RA (RAID5 Anonymous) meeting Enter: Andreas
Andreas: "Hi I'm Andreas and I have a RAID5 problem."
All: "Welcome Andreas."
272Zbigniew Proch0+1, simply smart, never again RAID5 :-)
273Robert YoungThere are many things in Computerdom, which exist simply because a lot of simple minded people choose to believe It is So.
XML databases is au courant. RAID/5 is an older version. Having a sufficiently trained scientific mind should be a requirement for doing databases; being able to separate the wheat from the chaff, as well.
274Marcin GuzowskiRAID5 or performance. The choice is obvious.
275Clinton AustinRAID 5 Sucks - period......
276Remigiusz BoguszewiczWell, I am on your side guys.
277Lars W. Andersen10 is actually more than twice as good as 5 ... amazing!!
278Nicolai Møller-AndersenFarewell ode to raid 5:
Everytime you go away -
it is good for harddisk space -
you really really make my day -
and make the world a better place
279Ejnar SvejstrupRaid-f sucks. It is an often-used sales trick to recommend raid-f in order to reduce the initial price on SAN appliances.
280Lars FohnThe video speech in CRN TV was convincing.
281Guy HarrisonI've been battling against RAID 5 since 1993. My 1997 SQL tuning book contained a section "Just say no to RAID 5"
282Arnold RimmerI'm invoking Spacecorps regulation 234 stroke 62.
283Floyd AbsherAs a DBA with over 10 years of experience, I think this is fantastic! Can we really purge the world of IT of ignorance...maybe not, but I'm still going to get a BAARF shirt and flaunt it.
284Bill MoranBecause RAID Free, Four, and Five suck
285Robert KlemmeI've read Art's article quite a while ago and was immediately convinced. I swear, I've never set up a RAID-5
286Bernd EckenfelsThere are simply too many reasons to not trust RAID-F. Learned it in Theory and Practice.
287Frank van BortelWho needs 5 in a binary world?
288Erik LundbyeRaid 5 is a low cost not a secure protection.
289Erik Andreas CayréNice learning from the masters;-) ..there will be no more RAIDF!
300Alan Hovgaard "Happytiger"Seeing is believing! But then how come so many ppl still believe in RAID5 seeing it fail all over. BAARF to the resque. So even the blind may see :-) .
301Aleksei SorokinMake RAID10, not War!
302Jaco VoslooIt's wonderful to see others feel the same! Please add RAID 6 (Fsix?) to the list.
303Henning RindbækNever seen a database perform really good on raid f... To the beancounters: Remember that diskspace is cheap, time is not.
304Glenn TrippSeems like a great idea! Why 14/2 shouldn't go down?
305Terry CrosbyRAID - that means Rage Against Indifferent Democrats, right???
306Thomas TeskeTake an overall-system perspective AND measure system behaviour. Then you have all you need - no need for debates about specific features, functions or designs. Message : Hello BAARF, master - what about BAARF videos on YouTube in English?
307Chris YungfleischDid you hear the one about the two database blocks stuck on Raid 5 disks ...
308Jon AdamsIt never fails to amaze me why the decision makers would believe sales people over their own in-house experts.
309Jurgen PlettinckxIt's hard to convince people that RAID 5 is not the right thing when looking for database performance ;-)
310Jens-Harald KnudsenAfter working more than 10 year's with Oracle, Sybase and MSSQL databases it's nice to know there still are people who looks at performance the same way I do.. Sadly my boss don't get it...
311Adam Tauno WilliamsViva la BAARF!
312Karol Korasadowiczabout time:-)
313Myron JohnsonI'm an IT consultant and have seen WAY too much data lost on RAID 5 arrays. RAID 1 or 10 for me, thank you
314Mischa E.J. HoogendoornFinally! We are free again!
315Jan AlfastsenAfter several performance issues on several DAS/SAN's, I have now proved my point about why one should not implement RAID3/5 to my CIO.
316Magnus JohanssonI have seen so many customers suffering from Raid5, sign me up please.
317Ben NobletJust lost my second (and last ever) RAID5 array to a dodgy Adaptec controller. No more RAID5 for me!
318Alexander 'sure' PodkopaevI'm pretty tired of customer's cry "We have good server, do we?" in case of lot of CPU power, some RAM and 3-5 spindles in RAID5....
319Louis AvramiOh, these UNIX admins and their love for RAID 5!
320Kim PedersenThe concept of RAID5 always made me feel uneasy, and mirroring seemed the way. I've met a lot of skepticism over the years, but having found BAARF have given me the great satisfaction of knowing I was right - SO TAKE IT!
321Selam AtevoliAwesome page. Excellent info on why RAID10 is the way to go if you truly give a crap about DB performance.
322Boris StefanovI've had enough of this *cheap* solution. Cheap to start with, very expensive on the long run. Kudos, BAARF
323Yass KhogalyHave had enough with RAID-5 over the past 22 years
324Ricardo NapoliRAID10 is twice as better than RAID5 (if not more)!!! Simple isn´t it ?
325Jeremy ColeRAID 10 forever.
326Anton SekerisAs a developer I am resposible for delivering high-performance, robust, hihgh-availability solutions in the financial industry. I have written an extensive document outlining how to configure RAID to work best with our applications. And then financial institutions use ... RAID-5! And then I have customer cases showing RAID-5 is not good. And then they still use RAID-5! And then they come up with test suites behaving in no way like our applications that 'prove' that RAID-5 is as good as RAID-10. I've had it! Death to RAID-5!
327keith littleSinging the BAARF song at KSC for many many years, and didn't even know I could join! Why do something good, when you can do something really, really stoopid with more hardware -- "...it's the way we've always done it...
328Daniel V. PedersenMmm .. Performance.
329Daniel MorganTo quote Edmund Burke: "The only thing necessary for the triumph of evil is for good men to do nothing."
330Henrik LaursenI just hate everything including RAID F
331Thomas RichterI BAARF
332Glenn FawcettHow many times must we fight this pointless battle?
333Sulman MansoorI am deeply impressed by the dedication of this GROUP towards this "HOLY MISSION" , if I may call it. I am in the process of convincing my Technical Administration to GET RID of RAID 5 which will need alot of convincing with proofs. Hopefully this forum will help me get through
334Mitchell MysliwiecI was duped into RAID-5, I'm writing this while waiting for a file to write
335Chris LeslieI have been tired of justifying my raid choices for years and am glad i found a place where people know the truth!
336Per KongstadI have over the years been fighting and explaining why this type of Raid is bad.
337Erik Bøg JensenBAARF
338Neels JordaanRaid 3,4,5 is evil.
339Kevin LittleDisks are cheap and manpower is expensive in relation. Why would anyone choose to save money on disks to introduce more human effort and chance for failure. Buy more disks, KISS, and LET ME SLEEP.
340Marc PerkelInteresting to find a group that came to the same conclusion I did. After experimenting it just seemed that Raid 10 because of simplicity and performance made more sense. I'm in the spam filtering business and Raid 5 seems almost as bad as the useless SPF crap we have to deal with.
341Brent WagnerI am joining BAARF because I am tired of impractical, ineffective, gee-whiz solutions to problems that don't exist.
342Tommy BollhoferHow many times must we fight this battle?
343Daniel Caballero RodríguezRAID5 has a lot of possible usages... DB usage is not one of them.
344Torolf GulstuenMy hair gets up every time I hear RAID 5. As a database administrator, I have always banned it.
345Torger EidemPreventing blaimstorming meetings.
346Steve CholertonHaving been led down the dark path that is Raid 5 d 5 in the past I have now seen the light. I am now fighting the good fight to rid the world of the evil that is Raid Free, Four or Five.
347Alex SonsTremendous!
As a Storage Architect having read the storageadvisor comments on RAID-5 reliability(*) and having encountered some real problems myself I can only agree with this initiative.
(*): http://storageadvisors.adaptec.com/2007/07/10/effect-of-drive-count-on-raid-5/
348Margeret McclaryI AM TIRED OF ARGUING
349Daniel ReyRAID FIX is the new evil!
350Jared HeckerGive me a RAID-5 array and I will give you an I/O-bound database...and it's so much fun to shove it in the storage manager's face...
351Guido van Versevelds!ck and tired of all those people that challenge that RAID-5 is not bad for DB environments.
352Graeme Galt3 failures of raid 5 on servers specd by muppets, countless hour of pain, server performance of a dead dog on a broken skateboard and im fed to goddamn up...
353Scott RogersRAID5 must die.
354Ben RogersBecause I can do basic arithmetic and basic arithmetic is not in RAID 5's favour.
355Jesper FrimannRAID5 is just another excuse to add complexity to save a few bucks on hardware.
356Ricardo Portilho ProniI won't comment anymore on RAID-F, not even for BAARF !
357Mark InmonTo find about RAID.
358Bryce ChidesterAt first it was just being frugal with disks... then it was just being suicidal.
359Ronald Asher HiggsI am a testing technician for a company that sells iSCSI SANs and virtual tape systems, almost exclusively configured with RAID5 storage. I thought RAID 5 was the ideal until I came across a mention of your group. I came. I read. I BAARFed.
360Martin BergerReplace the 'enterprise san' with JABOD, invest the money in memory and a well tested ASM.
361Matt DeMarcoI hate answering 'Why can't we use RAID 5...we have a lot of cache'
362Jim ParkWhen you need HA, there's no good reason to take a bat to I/O.
363Craig ThomasI'm all for anything involving beer and arguments with systems administrators.
364Lis ChaselingRumour has it that my IDS LUNs are about to magically morph from RAID10 into RAID5 - I think it must be beer o'clock *sigh*
365Graham KettlewellA catastrophic hardware raid on our main server that lost all the data on it. I made the mistake of thinking raid was a backup, not just a duty cycle tool. Never again!
366Russ PearsonRAID5 kills kittens! Stop it! Just say no to RAID5 for database implementations.
367Daniel HoffmanMy reason is clear, i was once lost in the Myst of lies about raid 5. how it was our only hope.. but i was lead astray by the false teachings. I have found the true like of Raid .. in 10
368Dinesh ChoudharyIn my career as a DBA I always have resisted against any quick solution without first understanding the problem. I found this forum rational and working towards busting the myth.It looks like that I found a forum to which I am compatible to.
369Mark CohenRAID-5 (3/4) is just plain stupid.
370Kenneth NaimTried of explaining that raid 5 won't perform, and will cost more money in the long run. And using it to test performance when production is raid 0+1 is worthless
371Evert MeulieYou make a good case, and I see no reason to disagree ;-)
372Don BauerMakes sense to me. Always wanted mirroring on my development system. Why play games?
373Terris LinenbachYou're right. I will never talk about RAID5 again!
374Henrik HarsfortStill too many RAID5 based OLTP-systems around
375Hari KaimalI've been administering Oracle databases (from 7.3.4 to 10gR2)on diverse platforms (AIX, Solaris, HP-UX, Win 2K, Win 2003, RHEL, RH9...) all of which have the worst thing in common: RAID-5. 'Nuff said.
376Peter HaynesRAID 5 is a slow antiquated technology. Nobody uses DOS systems anymore, they should not use RAID 5 either.
377Vasanth Kumar DReason for me being here is to know more about RAID
378Sam YapleDo people even know what they're getting into with RAID-F?!? I Couldn't sleep if I had data on a Five, Four, or Free.
379Karl FifeThanks, I needed that. Laughed out loud. Personally I don't like raid10 either, so maybe there should be a new club (or chapter) called BAARFT, (...Against Raid Five and Ten). I prefer straight mirrors or unstriped spans--slower but safer. Drives are usually fast and large enough now to forego the added risk of volume loss due to losing the 2 drives in the stripe. Remember, two DISMOUNTs (not just catestrophic mechanical failures) can cause total volume loss. On more than one occasion we've seen quirky drive/controller firmware interplay cause dismount during high I/O. This can mean losing the entire volume on rebuild, or when committing lots of transactions that are least likely to be backed up yet. I've also seen $h177y power and data cables cause dismounts resulting in dismout. That's Bad. Nothing insurmountable here, but still, it tips the risk reward analysis.
380Julio Cesar CorreaNo more RAID F*
381Charles BirdBAARF, RAID5 makes me sick
382Noel KoutlisRegards from Athens, Greece
383John NardelloI've lost multiple RAID-5 filesystems because of dual-disk failure during the rebuild stage. And management keeps buying them. Time to BAARF !
384Marcel Verdaasdon Nowadays we have false RAID controllers on Desktop systems. They are fun to play around with... Untill the OS crashes and you need to reinstall it totaly. Seems software RAID gives better perfomance still then a false controller. (Lucky me i used 1+0 in a software config)
385Brian Modra I once blindly followed "expert" advise to use RAID 5. Later I learned. But recently have been mis-advised again, by a major Telecom's data center administrator, that RAID 5 is the way to go. To convince him, from my humble perspective, I had to show him BAARF.
386James AllmondRaid-F is evil. DBA in 5 DB engines over the years, all perform poorly on Raid-F. Remember the old days, Raid 4 for this, 5 for this, 3 (God help us) for that...ugh, all it took was one person to add a file in the wrong place... RAID-F must die. RAID 1+0 is the way to go if you just have to use RAID... Anyhow disk really is inexpensive now..
387Brent WatkinsYo momma haz a RAID 5.
388Victor Pina Coutinho de JesusI read your explanation about raid5 and I completely agree about it. Nowadays it's no longer a choice.
389Glenn KelleyLast night Raid 10 saved my life and Raid 5 gave me a broken heart ... love thunderstorms
390Tristan HarmerBEFORE
3 disk R5 array. Database and web server performance graphs look like jagged, horrible mountains.

AFTER
6 disk R10. Graphs transformed to beautiful rolling fields. Bluebirds chirp, butterflies flutter and life is good.

R5?
NEVER AGAIN!!
391Waldemar Talen With high-performance terabyte drive prices affordable to the average family budget there is no excuse for sacrificing performance to "save space".
New technology is expensive to implement - RAID devices, i.e., SAN, NAS, etc. are not new and not expensive anymore - tuning around RAID 5 is expensive.
392Doug HannaI've inhereted a data center built by developers.
Every system's OS runs on a RAID 5.
I have a lot of work to do.
393Geoff StricklerI've spent far too much time arguing against RAID5 and for RAID10 (RAID1 or RAID0 for specific situations). I'm sick of arguing with idiots, and I'm thankful to have found BAARF.
394Marcus MönnigI've been there. I saw multiple hardware and software RAID-5 layers stacked up to hold database files. It turned out that "the application is slow".
395Romain BUnable to rebuild md0 ? WTF ???
396Naveen samalaThe title 'Enough is Enough'
397Marcin PrzepiorowskiTogether we can fight better with RAID-5 and big disks for DB. It drive me crazy every time I hear that SAN performance is almost infinity and every SAN can held every workload
398PollekeBecause i had a Linux software raid5 system crash on me...
399Scott Alan MillerBeyond performance problems, RAID-F also suffers from an additional level of failure not present in non-RAID-F (NRF) RAID systems like R1 and R10, parity failure. When a Raid-F (RF) array has detected a drive failure and begins rebuilding it is in a state of instability and an error on the controller can cause a complete array loss even in situations involving no drive failure and only involving momentary loss of contact with a drive. I have seen first hand entire RAID-5 arrays lost without the failure of a single drive and know other people who have gone through this as well.
Drives are cheap. It is time to move on to RAID 10!!
400Jerry LockeI have been bitten by RAID-5 failures and performance issues enough times to know better. Never again! Hardware is too cheap, vs. my time and frustration, for anything less than a mirror.
401Pablo MazzeiRAID5 is Bad.
RAID5 on your logfiles is worse.
RAID5 shared between many DB's logfiles is a living hell....
But the SAN has cache! They say.
BAARF FTW!
402Chip PurcellI'm about to walk into a Raid-5 environment. Your positive thoughts are appreciated. RETCH (Raid5 Erradication Team Curing Headaches) BAARF and RETCH?
403Robert PurdyHardly anyone realises the weakness of RAID 5 - they just see the potential performance hit and think I can live with that.
404Vic GennaWeed out the noise, and get an education.
405Jason DollerR-F is no longer practically viable.
406Kirk BrocasI have a deep aversion to RAID 5, both because it does not perform well in a database environment, and because it doesn't provide high enough redundancy.
It's a favourite of penny-pinching IT management and network administrators used to file and print servers.
Unfortunately DBAs are often the last people consulted when purchasing IT infrastructure hardware. As ASM grabs hold, the the roles of SAN administrator and DBA will collide, and this is the first front.
407Allen HerndonSA's are driving me crazy telling me raid 5 is cool now.
408Nayan MamaniaTo Share knwledge on RAID
409Dan JosephWoof!
410Al KalTwo logical write failure on RAID 5 on 2 drives at the same time caused full failure of Array...what is the point?
411Ansley BarnesRAID 5 is an obsolete tech that's getting more dangerous with age. Time to put it out to pasture!
412Steven VickruckHere's hoping I never have to hear about RAID-F ever again.
413Jake VinsonI lost a RAID 5 array and it cost over $13k to recover. I use only RAID 1 for personal data; RAID 5 in this office was a mistake. RAID 1 all the way unless you need 10 or 60 6 etc
414James PenneyWe do not support RAID-5; nor Malware for that matter...
415Paul van VelzenI have seen the light. No more discussions with the storage guys.
416Steven MoyseI was looking for information on RAID6 when I came across this page.
I have been fighting the good fight for many years now, but am losing the battle.
My customers machines are being replaced by VMs with SAN Storage. A vendor who supplied a RAID6 SAN system to one of my customers, he said that there was little performance impact when using RAID6.
I don't believe him, RAID6 sounds like RAID5 in sheep's clothing.
417Matthias BertschyAlmost lost 8T of data on a RAID5
418Diego Dutra SanchesI'm tired of bad database implementations, where the manager always ask to use RAID 5 to make solution cheaper
419Gregory Keith BrownEverything is a Transaction!
420Alex ReynoldsDrives are too cheap to risk your data on RAID 5.
421 Óðinn Burkni Helgasonerr... still using one RAID5 but that was from before... will change that asap...
422Dan FarrellPoor performance, poor reliability, why use RAID 5? Hard drives are cheap, RAID 10 rules!
423Pedro QuintasDont join in the dark side! Go BAARF!
424Michael M. HansenRAID 5 is BAD and EVIL! The main mail server at $WORK was down for 3 days because of a controller failure. The admin responsible had already been fired for incompetence, so we had no one to yell at.
We could only blame ourselves for not converting the array to RAID 1 at the earliest opportunity.
425Kenneth WitzellThe only thing worse than having your data and index files on RAID-5 is having them both on the same RAID-5 spindles!
426Brian KellyJust give me RAID 1+0, okay?
427Duke van LeeuwenWe all used raid 5....
Niels Bohr: An expert is someone who has made all the mistakes that can be made, but in a very narrow field
428Evgenij SmirnovI earn my living by doing things that make sense. RAID-F doesnt't make any. Period.
429Tom BascomI've had enough!
430Guzman BrasoGreat idea
431Reid McKinleyI see trees of green, RAID10 too...
I see it bloom for me and you...
and I think to myself,
what a wonderful world
432Thomas Vagnerraid 0+1 always did the job on my prods, never envountered problems. But i always see problems in raid5 confs in dev envs... tsk tsk
433Svend JensenRaid F optimizes space, something any accountant can relate to. Start talking about iops, you hit the wall of non understanding - and it is more expensive too..
434Franck HamonI can't support the performance of RAID5 and hate marketing of it !
435Brian WeissI have been sending our clients links to your site. They seem to just "love" the idea of RAID 5 - personally I can't stand it. Burned 3 times in the past 5 years by RAID 5. We do only mirroring with an imaging backup system (Acronis).
436Hans MolinSQL Server on RAID 5 sucks ass, no matter what I do.
437Markus BHate the RAID5...
438Vikram ThakurBetter late than never
439David CravenI've had enough. More than enough. Sadly, vendors will continue to sell this junk as long as they can find gullible, uninformed fools.
440Ronald L Copeland Sr.Will to know the truth...
441Abraham Gilles de JagerNo one in the managament believes RAID 5 is really poor in performance.
442Paul SeeberHi, finally I found you, brothers in spirit!
443Thomas WachBe consequent! RAID1 or RAID10, no reason for RAID3/4/5 anymore!
444Dave WalkerIf you don't trust one drive, why would you trust three or four?
445Parvez M MakhaniThanks for expanding our knowledge
446Bastian FeldmannJust had a server here with 3 slow disk in a RAID 5 array ... performed as expected: reading performance was 60% and writing performance 3% of a standard DELL server with 8 disks in a Raid 1+0 and 4 years old
447Hendrik VisageRAID-Z makes much more sense when you have bit rot!!
448Chad HembreeHello, my name is Chad and I am a recovering RAID 5 user. I openly admit the error of my ways and wish to be redeemed in the eyes of my fellow IT Managers and the DB Administrators we serve. I vow to to pickup the challenge and take on the fight against RAID F***.
449Keith PalmerThank you HP for yet another failed RAID-5 array. Death to RAID-F!
450Geoffrey ClineAlmost used Raid 5, thanks !!
451Ivan BajonI need a rifle, rations, and directions to the battlefield.
452Raphael BusschauRAID-F. You know F can stand for a lot more than Five, Four or Free.
453Ali FakoorWell got fed up with this [R]edundancy [A]ided [I]mplementation [D]isaster - [5] Take five
454Gustavo Machado GurgelOracle DBA
455Michael WebbSpindles are forever hidden from view. With all the cache and crappy code who needs redundant data spread out in pieces.
456Rich AllenTired of all the vendor misinformation.
457Brian BowmanConstantly battling (still) database performance disk bottlenecks with RAID5.
458Simon L. PrinslooIf commonsense was as common as the name implies and if disk space was cheap, everybody would use RAID10.
Oh, wait, disk space IS cheap but most of "them" always want RAID5? That implies....
459Julian MacasseyNice to know that my contempt for RAID 5 is shared. It's been lonely out there.
460Januaca TjandraOver 15 years I have been using RAID 5, experience many pains of data loss even during a time when scsi drive are the cost of 2 computers today and we build them in with two hot swap drive on stand by. I gave up on RAID 5 the moment hard drive reach 750GB at rubbish prices. RAID 10 is my preferred choice.
461Alexios KoumourisWorking for a company that has over 1000 oracle and sql installations on Greek Government. Before 8 years, most of our clients used raid-5. Now, most of them use Raid10 or raid 1.
462Jan SchermerI hate RAID5 wholeheartedly - people think it's "safe" to use with "good" performance - the exact opposite is true.
463Stork HsuI'd like to learn from others experiences in planning SAN for Oracle environments.
464Alexandru DanI am so TIRED of people arguing about RAID-F and why is better or why would they choose it. I would never choose it and never done it. I've done RAID-1, RAID-0 and RAID-10 but never any RAID-F's.
465David BridgesI've seen the performance impact of Raid5 rebuilds after drive failures. Also, the issue of building drive arrays from a single prod run of drives of the same age (multi drive failures)
466Luis BognotgOOD INFORMATION
467Gary J. HayersRAID-5 seemed like a good idea at the time...
468Antony RuddyI totally concur with BAARF's philosophy.
469Frank Rotchford In 1992 I refuted the RAID-5 claims, but no one would listen. This 40 year DBA now feels vindicated.
470Art TaylorRAID-F creates a false sense of security. It's faster to restore from backup than it is to rebuild an array. If we just admit that, then mirroring = faster MTTR. Spindles and chips are less expensive than effectively being out of business for days.
99999Keith MooreCurrent employer uses RAID 5 SAN. Purchased a new "faster" SAN. With minimal load, read performance increased 4X. Write performance exactly the same. Hmmmmmmmmmmm!

This page was last updated on 17 Feb 2007.  Please mail comments, critics and ideas to WebMaster@MiracleAS.dk.