Oracle FAQ Your Portal to the Oracle Knowledge Grid
HOME | ASK QUESTION | ADD INFO | SEARCH | E-MAIL US
 

Home -> Community -> Mailing Lists -> Oracle-L -> RE: FW: SAN configuration for Banner

RE: FW: SAN configuration for Banner

From: Paul Drake <discgolfdba_at_yahoo.com>
Date: Thu, 20 Nov 2003 18:05:00 -0800
Message-ID: <F001.005D74E4.20031120180500@fatcity.com>


"Above all else, you shall do no harm".  

I never have pulled a live drive, although I have failed drives via the management software, downed a path to a FibreChannel switch to test multipathing, pulled an ethernet cable to test trunking/adaptive fault tolerance/adaptive load balancing, downed a listener to test connect-time failover.  

Its one thing to see failover work at a vendor's booth, its another thing to see it working on a system that didn't necessarily have the best drivers configured by Intel's best techs.  

Its just fun thinking about the vendors' horror. :D  

Pd

Matthew Zito <mzito_at_gridapp.com> wrote:  

One of the issues in this situation is that the drive sleds usually have circuitry in them to insure electrical isolation and that once a disk is marked as dead, the disk is spun down and kept from placing any traffic onto the backend bus. When a disk is just pulled, problems can happen, moreso on the old SCSI bus architectures than Fibre, but its been known to happen. That's why many storage vendors make "fail disk X" commands, to simulate disk failures without physically yanking a sled.  

Heh - I just saw your post, Jared - very kind of you to say.  

There's a lot of good traffic already on this subject - on some arrays the only way to do proper RAID-10 is by making a bunch of RAID-1 volumes on the array side and doing software striping. As far as the reasons vendors don't want to push 0+1, its really a matter of being able to walk into a boardroom and tell a customer "our array can support up to (insert absurd amount of space here) in one frame". What they don't say is that in order to do that you need to use big slow 7200 RPM drives and huge RAID-5 groups. It's marketing talk. Also, vendors like to offer that to C-level folks, who then feel like they're being responsible by not buying "expensive" storage.  

There's a great whitepaper that EMC has on RAID-5 optimizations on the Clariion CX-series if you're forced to go that route. Get your Dell engineer to poke someone at EMC to get it - I don't have it anymore (I wish I did).  

Thanks,
Matt

--

Matthew Zito
GridApp Systems
Email: mzito_at_gridapp.com
Cell: 646-220-3551
Phone: 212-358-8211 x 359
http://www.gridapp.com

-----Original Message-----

Sent: Thursday, November 20, 2003 7:44 AM To: Multiple recipients of list ORACLE-L

Hmm  

The small print on compaq SANS specifically states that you shouldn't pull drives 'for testing purposes' but only do it in the event of a drive failure. I wonder if other vendors say the same thing to avoid exactly this stuff.  

Mind you however I think of it I still cannot understand why SAN vendors don't insist on RAID10 anyway, more disks,more cache,more controllers=more cash. Sales and Marketing must be a strange world.  

Niall

-----Original Message-----

Sent: 20 November 2003 07:05
To: Multiple recipients of list ORACLE-L

Jared,  

I would have preferred RAID 10, and that is what I has asked for, but did not apparently receive. I've lost RAID 01 volumes due to a failure of 2 drives, where a RAID 10 volume would have survived.  

This reminds me of a crowded flight I took where I read reading an OCP study guide. The 2 gentlemen on either side of me were both database/architect consultants. The war stories soon followed. The gent on the left talked about a project, where the vendor was supposed to have provided RAID 10. When everyone had agreed that everything on the checklists had been completed and all were ready to sign-off, he pulled 2 drives, on separate trays, not in the same physical location. If the config was RAID 10 (as per spec), it would have sailed along. It crashed.  

Damn, someday I hope to have enough testosterone to pull that off. I would love to see a vendor's facial expression when a system ready to go live is downed, due to their arrogance that no one would call their bluff.  

This technique would also turn up a RAID-5 system as unworthy. I would see hot-sparing a RAID-0 stripe (half a RAID 01) as mandatory.  

Paul

Jared.Still_at_radisys.com wrote:

Paul, just curious why RAID 01 rather than RAID 10.

RAID 10 is much more resistant to disk failures that RAID 01.

10 drives in a RAID 01 with drives having a MTBF of say 100,000 hours. ( don't know if the 100k is low or high )

Each stripe in the RAID 01 has a MTBF of 20k hours, which is 10k hours for the array as a whole.

In a RAID 10 each mirrored pair would have an MTBF of 50k hours, which also appears to give an MTBF of 10k hours, but this is where statistics are misleading.

Lose one drive on each side of the RAID01 and is down.

The RAID 10 could lose 5 up to drives, as long as each failure is only one side of a mirrored pair, and still stay up.

Even more beneficial, in the RAID 01 if you lose on drive, you will lose 50% of read throughput.

Losing a single drive in the RAID 10 will cost you about 10% throughput.

If Matthew Zito is still here, he can no doubt enlighten us with no end of detail on the subject.

Jared

Paul Drake <discgolfdba_at_yahoo.com>
Sent by: ml-errors_at_fatcity.com
 11/19/2003 12:20 PM
 Please respond to ORACLE-L

        
        To:        Multiple recipients of list ORACLE-L <ORACLE-L_at_fatcity.com> 
        cc:         
        Subject:        Re: FW: SAN configuration for Banner


Sam,   

Dell was perfectly happy to configure the following for me on a CX200 unit: (2 trays of 15 drives, 1 tray 36 GB 15K, 1 tray 73 GB 10K, 1 hot spare per tray)   

3 x RAID 1 for online redo logs (members ! a,b,c, 6 drives total) 2 x RAID 01 for datafiles (8 drives each, 16 drives total) 1 x RAID 1 for archlogs (2 drives)
1 x RAID 01 for undo (4 drives)   

we used less than half of the available storage, saving that for when another system is migrated to this unit. when creating filesystems, we did not format the entire RAID vols for live files, and created filesystems for storing local backup sets.   

You will like having datafiles on 8 drive RAID 01 (or RAID 10) volumes. RMAN backups fly like you would not believe.   

hth.   

Paul

Sam Bootsma <sbootsma_at_gbrownc.on.ca> wrote: Hi List,   

We are approaching the cusp of a decision on how to store Oracle data files on our SAN. We don't have the SAN yet, but it is due to arrive any week (if not any day).   

I passed Cary's "Is RAID 5 Really a Bargain?" paper to our Sys Admin, which he read and succinctly summarized for the Technical Manager here. I have also read through a couple of papers referenced in the BAARF site. The Sys Admin comments were:   

Dell would like to know what RAID mode we want configured on the SAN for the B80 and 6C4 computers. Sam has told me that, in the Oracle community, mirroring (RAID1) is preferred over RAID 5 for various reasons (RAID5 is: more costly for write-intensive applications, 3 times more likely to incur data loss, suffers from massive performance degradation during partial outages). RAID1 will be more costly per unit of usable storage. Mirroring seems to be the best choice. Let me know what you think.   

Here is the Manager's response:   

Any suggestions on how I can counter points 4 and 5 and the last point before his "Thanks" line? Currently we have two B80's (AIX 4.3.3) set up in a HA configuration. They share an external disk array. So if a hardware compo! nent in the primary box fails, then it will automatically failover to the secondary box (and at the same time, the secondary box takes control of the external disk array). I think the clustering term in point (4) is referring to this setup.   

Thanks for any suggestions.   

Sam.      

Sent: November 18, 2003 5:08 PM   

All the points are valid...however..my thought processes were as follows:   

  1. The System & Core Application disks are resident on the disks within the CPU and Mirrorred (Everyone OK with that I think)
  2. The Databases are Resident on the SAN
  3. The SAN disks are RAID 5 as the provide more usable space for the cost as compared to mirrorring
  4. As the IBM Systems (B80's & 6C4's) are clusterd thus effectivley Mirrors the RAID 5 Arrays mitigating the issues Sam raises re preformance degradation (which will only ever arise in the event of a failed disk/automated rebuild which is usually configurable to address performance degradation)
  5. Write to Disk/Commit to Database should be a background process (although I recognise this is a transaction/write intensive based system)

This is a standard model that all servers are being deployed with and unless there are any specific technical reasons why this will not work it is the way I would like to see the systems implemented. Remeber, with the SAN...Reconfiguration of Disks is not a large issue anymore if required in the future.   

Although not an AIX/Oracle guy...I disagree with the statement that RAID5 is 3 times more susceptable to incur Data Loss. RAID 5 is a proven technology   

Thanks..... Andrew   

-----Original Message-----

Sent: Tue 18/11/2003 2:56 PM
To: Andrew Riem   

Dell would like to know what RAID mode we want configured on the SAN for the B80 and 6C4 computers. Sam has told me that, in the Oracle community, mirroring (RAID1) is preferred over RAID 5 for various reasons (RAID5 is: more costly for write-intensive applications, 3 times more likely to incur data loss, suffers from massive performance degradation during partial outages). RAID1 will be more costly per unit of usable storage. Mirroring seems to be the best choice. Let me know what you think.   

Carl




Do you Yahoo!?
Free Pop-Up Blocker - Get it now
--

Please see the official ORACLE-L FAQ: http://www.orafaq.net
--

Author: Paul Drake
  INET: discgolfdba_at_yahoo.com
Fat City Network Services    -- 858-538-5051 http://www.fatcity.com
San Diego, California        -- Mailing list and web hosting services

---------------------------------------------------------------------
To REMOVE yourself from this mailing list, send an E-Mail message
to: ListGuru_at_fatcity.com (note EXACT spelling of 'ListGuru') and in the message BODY, include a line containing: UNSUB ORACLE-L (or the name of mailing list you want to be removed from). You may also send the HELP command for other information (like subscribing). Received on Thu Nov 20 2003 - 20:05:00 CST

Original text of this message

HOME | ASK QUESTION | ADD INFO | SEARCH | E-MAIL US