toasters

toasters@lists.teaparty.net

13515 discussions

F760 Cluster question when using ATM
by David Lockwood 29 Jun '01

29 Jun '01

I am trying to work out how to configure an F760 cluster and appear to be running out of slots. The main connection from the user community to the filer heads is via ATM, for a cluster I am told I need 2 ATM cards for each head in an active passive configuration, the passive connection taking on the identity of the 'failed' (heaven forbid) filer. Now, due to the number of disks required for data storage I need two FCAL loops per filer, 4 required in each head for failover. I need 1x 100Mbit ethernet connection to connect directly to a second cluster for mirroring and 2x 100Mbit direct connections to Unix servers for Oracle database access. The F760 has 9 slots. I can use slots as follows: Slot 1 Quad (or dual ethernet) Slots 2 and 3 for ATM Slots 4,6,7 for FCAL cards Slot 8 for Cluster Card Slot 9 for NVRAM It appears I cannot use Slot 5 for any type of card other than a connection to a tape device! Can I fail the ATM connection over to a 100Mbit ethernet slot, if I use a quad ethernet card then there would be a connection free. This would free up Slot 3 for the fourth FCAL card. Does anyone have any ideas on this and how we can get it all to fit. Before anyone suggests it the answer is no I cannot look at the 840's just yet! Thanks in advance David ---------------------------------------------------------------------- For further information on Benetton Formula 1 visit our web site at www.benettonf1.com. This email and any files transmitted with it are confidential and intended solely for the use of the individual or entity to whom they are addressed. If you have received this email in error please destroy and notify Benetton Formula on +44 1608678000." ----------------------------------------------------------------------

1 0

RE: Oracle on NetApp - opinions/experiences requested
by Neil Stichbury 29 Jun '01

29 Jun '01

I'm one of a small team that supports a SUN6500 environment with Oracle 7 on a 760 Filer. The interface is dual Gigabit Ethernet fiber using round-robin. The cards are SBus and although performance is good we intend fitting PCI cards. Solaris 2.6 and 6.01 on the filer. Initially there were performance issues but the logfiles were moved to local disk (18GB mirror), Oracle tuned a bit and it goes like a rocket. Supporting approx 1400 users. Snapshots with hotbackup mean that availability is almost 24x7. Netapp whitepapers on 'Oracle on the filer' are a good starting point. Can't recall any 'negative' experiences actually !. Regards, Neil Stichbury gen-i Limited New Zealand -----Original Message----- From: Jeff Kennedy [mailto:jlkennedy@amcc.com] Sent: Saturday, 23 June 2001 07:38 To: NDMP List Subject: Oracle on NetApp - opinions/experiences requested Of those who run Oracle on Netapp, how do you feel about it? I recently received an e-mail in which a Sun rep stated they gave up running Oracle on NetApp over a year ago. "Gave up" implies there were issues that were not easily overcome. According to NetApp, Ontap is "really a database file system". The technical specs on Ontap seem to back this up, but many people shy away from the idea. Granted, most of the people knocking the idea are other storage vendors or VAR's but some concern in the admin community I've spoken with revolve around NetApp being NAS and not DAFS or SAN. I'm looking for experiences, both positive and negative, from anyone doing this. Thanks. -- ===================== Jeff Kennedy Unix Administrator AMCC jlkennedy(a)amcc.com This e-mail message has been scanned and cleared by MailMarshal http://www.gen-i.co.nz This e-mail message has been scanned and cleared by MailMarshal http://www.gen-i.co.nz

2 1

Re: [Veritas-bu] NDMP Restores
by W. Curtis Preston 28 Jun '01

28 Jun '01

Grant, Thanks for the information. I'd like to comment on a few things. First, let me say that I am a fan of NDMP, as it is giving us a valid solution for backing up all these NetApps I keep seeing everywhere. ;) Before valid NDMP server and client software was in place, all we had was NFS. I have a lot of respect for NetApp and the other backup and recovery options you have (Snapshots, SnapMirror, etc.), but I would like this one issue addressed. At 11:45 AM 6/27/2001 -0700, Melvin, Grant wrote: > NetApp's dump is no stranger to this, we have added support > for qtree metadata, NT ACLs, Unicode, NT Streams. However, a > standard BSD restore compatible application such as Solaris' > ufsrestore can still restore the data, but will only restore > NFS attributes. So should you have zero filers in 10 years > (heaven forbid) and a pile of NDMP/dump tapes to restore from > then as long as you have access to a dump compatible restore > program you should be able to manually retrieve your data. The key for the original poster to understand is the word _manually_. This means that you can load the tape into a Solaris box (or any box that can read BSD dump, such as HP, AIX, Linux), do some manual tape manipulation, and be able to restore the data (without the NT-specific info). This is a far cry from being able to point and click your restores into existence. > Legato's BudTool product proved that NetApp data could be > restored on non-NetApp storage either in a disaster recovery > scenario or because that is what the customer wants. We'd like > to think that some if not all of the current 7 NDMP-compliant > backup software applications that support NetApp could add the > same capability. Yes, they could, and I wish they would. I wonder if some of us should just write a tool that does that? Hmmm.. Gears churning.... I wonder if we could figure out a way to feed these tapes to AMANDA, and have AMANDA automate the restores. Or just write our own utility... Hmm.... Anyone interested in this project? > Marion asks for a standard backup tape format that can be > interchanged between NDMP-compliant vendors and we applaud > her for it. However, that is beyond the current scope of the > NDMP community or the SNIA Backup working group in general. > > We think its going to take an awful lot of persuading to get > all the current backup software vendors to define a single > format and then have all the possible system vendors support > it. Most companies have chosen their solution based on > trying to differentiate themselves from all the others and > we see that continuing for some time. Yes, it would take some doing to accomplish that. However, we've got to start somewhere. My biggest issue with NDMP is that it allows the OS vendor (in this case NetApp) to determine the backup format. I understand that it would be a lot of work to get this accomplished, but I don't think it's outside the realm of possibility. Trying to differentiate yourself in the backup world is a bad thing. (I submit for your perusal IRIX's XFSDUMP.) I'm glad to hear that NetApps tapes can at least be read by a dump-knowledgeable box. But you're not the only vendor out there. Yes, I know that it says right in the V5 spec that backup format is out of scope. Yes, I know that getting a bunch of vendors to get together is not an easy thing. But I'll keep saying that this should be in scope until somebody listens: OS-SPECIFIC BACKUP COMMANDS BAD I do wish that Veritas would make their multiplexed backup format a little more open, and make a command that can easily read a NetBackup multiplexed tape if NetBackup is dead, but at least their backup format is the same on every box. The idea of going back to the days when I had Solaris, HP, and Ultrix dump tapes, and 3B2 cpio tapes, and VAX backup tapes gives me chills. And, given the upcoming prominence of the NAS industry, we may have just that. (Assuming you're in a shop that has more than one NAS vendor.) That would stink. Anybody that agree with me, feel free to send a "me too" message. --- W. Curtis Preston Principal Consultant for Storage Designs, your storage experts Webmaster: http://www.backupcentral.com Phone: 760 653 1007

3 2

RE: Vol copy max speed ? alternatives ?
by Mohler, Jeff 27 Jun '01

27 Jun '01

Myself..I havent seen col copy/ndmpcopy/snapmirror do much over 18-22Mb/sec depending on platform. If I had a free hour Id love to go into my Lab and prove myself wrong, but I wont be able to until at least tomorrow if anyone is interested in some data between some 880s. -----Original Message----- From: Fox, Adam [mailto:Adam.Fox@netapp.com] Sent: Wednesday, June 27, 2001 1:46 PM To: 'FPS France'; toasters(a)mathworks.com Subject: RE: Vol copy max speed ? alternatives ? I guess I'm not sure what a 'perfect' 10BT connection speed is and so the comparisons confuse me. By my calculations: 10Mb 1MB 1GB 60s 60 min ---- ------- ------ ------- --------- = 4.39 GB/hr. 1s 8Mb 1024MB 1 min 1 hr This is 10BT at it's maximum throughput...not counting for protocols, etc. So let's look at 36 GB/hour using the same method: 36GB 1024 MB 8Mb 1 hr 1 min ----- ------------ ------ ------- ---------- = 81.92 Mb/s 1 hr 1 GB 1MB 60 min 60 sec So you're a little under a "perfect" 100BT from my calcuations. Not great, but not as bad as I read it to be..and 18X better than 10BT. If the purpose is to reduce downtime (perfectly resonable) then perhaps you need something that can do incremental changes instead of doing it in one shot. ndmpcopy can help. It's free and has incremental support even if it's a little slower, if you can reduce the size of your last incremental, it won't hurt as much if the data you are trying to move while you are down is smaller. SnapMirror is *great* for this. The only downside is it is not free. You'll need to work out a deal with your rep if you feel this is really the way you want to go. Just some thoughts. -- Adam Fox NetApp Professional Services, NC adamfox(a)netapp.com > -----Original Message----- > From: FPS France [mailto:fpsfrance@fps.fr] > Sent: Thursday, June 21, 2001 11:20 AM > To: toasters(a)mathworks.com > Subject: Vol copy max speed ? alternatives ? > > > Hi everyone, > > We are looking for a high rate migration operation which > could let us move a > volume from a disk shelf to another. > In a first place, we think of using the vol copy command. We > thaught we > could easily manage a strong volume copy rate but it appeared > the rate is > buggy limited to 36Gb/hour (as if the Filer was copyinbg over > a 'perfect' > 10Bas-T interface) : > http://now.netapp.com/Knowledgebase/solutionarea.asp?id=3.0.72 95526.2875209& resource= In looking around we found papers dealing with others interface kinds : http://now.netapp.com/Knowledgebase/solutionarea.asp?id=3.0.7295507.2875209 The limit are 25GB/hr over 100Base-T interface, and 28GB/hr over FDDI so less than a 'perfect' 10Base-T interface. (on a F630 Filer :) Reasons given by the NetApp support is the CPU has a bunch of work in calculating optimizations of read and write operations. But I would be sure we are really not strunglled by any other limitations. So here are our two questions : - (just to be sure) did anyone succed in using the vol copy command once and go beyond these limits ? -what is your opinion about that alternative : plug the Filer on a private Giga-ethernet network and make a manual copy (copy -R mnt/nac/vol0 /other_place ) of the whole volume from a shelf to the other. Do you think we could obtain real high rate transfers while keeping the integrity of the volume ? (the volume is a root volume with the "/etc" directory), no special tricks needed (if I use Windows, may I expect ACL modifications) ? I thank you in advance.

1 0

NAS Benchmarking
by Robert Mance 27 Jun '01

27 Jun '01

What are some of the favorite methods of benchmarking a NAS? -- Robert Mance robert(a)cinesite.com Manager of Technical Operations Cinesite Digital Studios 323-468-5711 (Work) 1017 N Las Palmas Ave., Suite 300 323-468-2180 (FAX) Hollywood, CA. 90038 213-391-7527 (Pager)

1 0

RE: Vol copy max speed ? alternatives ?
by Fox, Adam 27 Jun '01

27 Jun '01

I guess I'm not sure what a 'perfect' 10BT connection speed is and so the comparisons confuse me. By my calculations: 10Mb 1MB 1GB 60s 60 min ---- ------- ------ ------- --------- = 4.39 GB/hr. 1s 8Mb 1024MB 1 min 1 hr This is 10BT at it's maximum throughput...not counting for protocols, etc. So let's look at 36 GB/hour using the same method: 36GB 1024 MB 8Mb 1 hr 1 min ----- ------------ ------ ------- ---------- = 81.92 Mb/s 1 hr 1 GB 1MB 60 min 60 sec So you're a little under a "perfect" 100BT from my calcuations. Not great, but not as bad as I read it to be..and 18X better than 10BT. If the purpose is to reduce downtime (perfectly resonable) then perhaps you need something that can do incremental changes instead of doing it in one shot. ndmpcopy can help. It's free and has incremental support even if it's a little slower, if you can reduce the size of your last incremental, it won't hurt as much if the data you are trying to move while you are down is smaller. SnapMirror is *great* for this. The only downside is it is not free. You'll need to work out a deal with your rep if you feel this is really the way you want to go. Just some thoughts. -- Adam Fox NetApp Professional Services, NC adamfox(a)netapp.com > -----Original Message----- > From: FPS France [mailto:fpsfrance@fps.fr] > Sent: Thursday, June 21, 2001 11:20 AM > To: toasters(a)mathworks.com > Subject: Vol copy max speed ? alternatives ? > > > Hi everyone, > > We are looking for a high rate migration operation which > could let us move a > volume from a disk shelf to another. > In a first place, we think of using the vol copy command. We > thaught we > could easily manage a strong volume copy rate but it appeared > the rate is > buggy limited to 36Gb/hour (as if the Filer was copyinbg over > a 'perfect' > 10Bas-T interface) : > http://now.netapp.com/Knowledgebase/solutionarea.asp?id=3.0.72 95526.2875209& resource= In looking around we found papers dealing with others interface kinds : http://now.netapp.com/Knowledgebase/solutionarea.asp?id=3.0.7295507.2875209 The limit are 25GB/hr over 100Base-T interface, and 28GB/hr over FDDI so less than a 'perfect' 10Base-T interface. (on a F630 Filer :) Reasons given by the NetApp support is the CPU has a bunch of work in calculating optimizations of read and write operations. But I would be sure we are really not strunglled by any other limitations. So here are our two questions : - (just to be sure) did anyone succed in using the vol copy command once and go beyond these limits ? -what is your opinion about that alternative : plug the Filer on a private Giga-ethernet network and make a manual copy (copy -R mnt/nac/vol0 /other_place ) of the whole volume from a shelf to the other. Do you think we could obtain real high rate transfers while keeping the integrity of the volume ? (the volume is a root volume with the "/etc" directory), no special tricks needed (if I use Windows, may I expect ACL modifications) ? I thank you in advance.

1 0

NDMP Restores
by Melvin, Grant 27 Jun '01

27 Jun '01

Hi, Yesterday there was some discussion on these mailing lists regarding NDMP and limitations on restores, etc. We'd like to thank you for your feedback and to explain what the current focus of NDMP is and in particular how NetApp's chosen long term archivable backup format (dump) works within NDMP. Over the last few years NDMP has grown from essentially a 2 company (PDC then IntelliGuard now Legato and Network Appliance) backup solution to a community with 20+ companies participating and > 25 NDMP compliant products shipping. The community has backup software companies, tape library vendors and file system vendors working together to achieve true interoperability and a growing set of features. This was the vision of the original companies. Recently, this has resulted in a new V4 specification that essentially is a cleanup of V2 & V3 and has added extensibility so that new features may be added without waiting for a spec change. V4 implementations will arrive this year from multiple companies and should put an end to interoperability issues that have cropped up from time to time. NDMP (http://www.ndmp.org) is essentially a "plumbing" api for controlling data protection/management operations right now and has remained data format independant/neutral. This means that it is relatively easy for backup software companies to "plumb" a backup or restore that uses the native format(s) of each system vendor. There are 3 basic data formats commonly used in tape backups, cpio, tar & dump (should we include MicroSoft tape format?). Some operating systems such as Solaris currently support all three. However, some systems such as WAFL support one, i.e. BSD's dump. Even though systems vendors have these native commands, most backup software companies choose to write their own agents based on one of the 3 formats discussed and tend to add some "proprietaryness" as they support more and more dissimilar file systems or add performance optimizations such as multiplexing, etc. NetApp's dump is no stranger to this, we have added support for qtree metadata, NT ACLs, Unicode, NT Streams. However, a standard BSD restore compatible application such as Solaris' ufsrestore can still restore the data, but will only restore NFS attributes. So should you have zero filers in 10 years (heaven forbid) and a pile of NDMP/dump tapes to restore from then as long as you have access to a dump compatible restore program you should be able to manually retrieve your data. Legato's BudTool product proved that NetApp data could be restored on non-NetApp storage either in a disaster recovery scenario or because that is what the customer wants. We'd like to think that some if not all of the current 7 NDMP-compliant backup software applications that support NetApp could add the same capability. Marion asks for a standard backup tape format that can be interchanged between NDMP-compliant vendors and we applaud her for it. However, that is beyond the current scope of the NDMP community or the SNIA Backup working group in general. We think its going to take an awful lot of persuading to get all the current backup software vendors to define a single format and then have all the possible system vendors support it. Most companies have chosen their solution based on trying to differentiate themselves from all the others and we see that continuing for some time. Cheers, Grant and Greg Greg Linn Manager, NDMP Development linn(a)netapp.com 408.822.3752 =========== grant(a)netapp.com ========== Grant Melvin === === Software Development Manager === === Data Availability & Management === |\ | __ ___ /\ __ __ === Network Appliance === | \ | |__ | /__\ |__| |__| === 475 East Java Drive === | \| |__ | / \ | | (R) === Sunnyvale === === California, 94089 === === Tel:(408)822-6761 =========== Network Appliance ========== Fax:(408)822-4578

1 0

Migrating from one NETBIOS name to another
by Brian Tao 26 Jun '01

26 Jun '01

As a result of some rushed planning, all of our Windows desktops and any NT/2000 applications that need to access one of our filers do so using the UNC path "\\na1". We would like to change that to "\\corp-na1" for a variety of reasons. I know about the cifs.netbios_aliases option, so the filer will answer to both "\\na1" and "\\corp-na1". We can change our NT login scripts so the default mappings are updated, but I'm worried that there may be untold number of other applications, saved path lists, config files, etc. still referencing "\\na1". Is there a way on the filer or on the PDC to log CIFS sessions attaching to one NETBIOS name vs. the other, and the source of that connection? I'd like to know if we are faced with hundreds of users and processes still using the old name (which will break when we take it away), or just a few here and there. I could resort to doing pktt's on the filer and picking apart netbios packets, but I'd rather avoid that if possible. ;-) -- Brian Tao (BT300, taob(a)risc.org) "Though this be madness, yet there is method in't"

1 0

RE: [Veritas-bu] NDMP restores
by Michael Owen 26 Jun '01

26 Jun '01

Yes we are listening. Some "other" systems do use non proprietary backup implementations with standard TAR formats. Unfortunately the backup companies dictate the format for enterprise level backup solutions. NDMP is not a real solution for enterprise level backups, but it the only implementation currently available for NAS systems. MO -----Original Message----- From: owner-toasters(a)mathworks.com [mailto:owner-toasters@mathworks.com]On Behalf Of Marion Hakanson Sent: Tuesday, June 26, 2001 1:14 PM To: Karen Schoenbauer Cc: Fayek, Ahmed; veritas-bu(a)mailman.eng.auburn.edu; 'dl-ndmp-users(a)mentor.com'; toasters(a)mathworks.com Subject: Re: [Veritas-bu] NDMP restores Karen wrote: > NDMP restoresNDMP restores can only be restored to the same NAS vendor type. > So if I did an NDMP backup to a NAS vendor box called NetApp1, I can run a > restore to NetApp2 as longs both boxes are NetApps. You can not restore a > NDMP backup to a solaris box or to a different NAS vendor. Actually, that doesn't hold true for all NAS systems. With BudTool, one can restore NetApp NDMP backups to a Solaris host using "ufsrestore", although you lose the CIFS (NT) ACL's. I've done this on several occasions, and it's a side benefit of BudTool's philosophy of using the OS/vendor native backup utility and writing that image to tape. Yes, you do need the vendor's software to do a restore, but you don't even need BudTool, if you're sufficiently prepared. Personally, I'd rather see NAS systems adopt something like the relatively accessible POSIX-tar-compatible format that QuickRestore uses. Then one could use GNU tar or "pax" as a fall-back, and one would have a lot less trouble with the problem of long-term archival. Anybody out there listening? Regards, -- Marion Hakanson <hakanson(a)cse.ogi.edu> CSE Computing Facilities

1 0

restores again
by Fayek, Ahmed 26 Jun '01

26 Jun '01

Hi all where can i find a log to tell me which files have been restored, from where, to where and when ? Regards Ahmed

2 1

Jump to page:

2025

2024

2023

2022

2021

2020

2019

2018

2017

2016

2015

2014

2013

2012

2011

2010

2009

2008

2007

2006

2005

2004

2003

2002

2001

2000

1999

1998

1997

toasters