Guys,
We're looking to spec out a big Netapp cluster, and since 7-mode is end of life in terms of new features, and it looks like 8.2.1RC1 is now out, what do people think of it?
One question I have is whether I should get a bunch of 3250s in cluster pairs, on the order of four or six heads all clustered together, or whether it makes more sense to get a 6250 pair (or pairs?) instead.
We're looking at roughtly 1 Petabyte of storage. Probably 50% can be SATA, the rest SAS, but possibly <20% would need to be FC LUNs for Oracle DBs doing some ERP stuff. Lots of IOPs will happen there. Another thought is to get SSDs and setup FlexPools. Which is nice since you can prioritze volumes to get that speedup. But which costs like crazy from what I hear.
I'm hesitant of putting all my eggs into one big basket.
After talking with a FE, it looks like there are all sorts of goodies in 8.2.1 and up, such that we can setup multiple Virtual Filers (now known as SVMs I think) on the cluster and do transparent volume moves across heads/aggregates, etc. And moving a SVM from one cluster head to another or to another set of nodes in the cluster.
My question of course, is how well dows this work in practise? Esp for FCP luns. Being able to live move volumes between aggregates looks to be a real help, since now I won't be stuck withe large, but not endless aggregates.
Basically, I'm looking for what people think of how well this all works in real life and what the gotchas are. And of course, what about tape backups over direct attached SAN FC tape drives from an SVM? How is that performance as well?
Thanks, John
I use volume moves all the time.
It works seemlessly. Have not tried the LUN move yet. No plans to. I distrubted them across the heads anyway.
The 3250's will likely be good. If you expect to pound the crap out of your systems like I do, you might want to shoot for the 6000 series. More RAM, dedicated NVRAM more processors, etc.
--tmac
*Tim McCarthy* *Principal Consultant* =
Clustered ONTAP Clustered ONTAP NCDA ID: XK7R3GEKC1QQ2LVD RHCE6 110-107-141https://www.redhat.com/wapps/training/certification/verify.html?certNumber=110-107-141&isSearch=False&verify=Verify NCSIE ID: C14QPHE21FR4YWD4 Expires: 08 November 2014 Current until Aug 02, 2016 Expires: 08 November 2014
On Thu, Dec 19, 2013 at 5:14 PM, John Stoffel john@stoffel.org wrote:
Guys,
We're looking to spec out a big Netapp cluster, and since 7-mode is end of life in terms of new features, and it looks like 8.2.1RC1 is now out, what do people think of it?
One question I have is whether I should get a bunch of 3250s in cluster pairs, on the order of four or six heads all clustered together, or whether it makes more sense to get a 6250 pair (or pairs?) instead.
We're looking at roughtly 1 Petabyte of storage. Probably 50% can be SATA, the rest SAS, but possibly <20% would need to be FC LUNs for Oracle DBs doing some ERP stuff. Lots of IOPs will happen there. Another thought is to get SSDs and setup FlexPools. Which is nice since you can prioritze volumes to get that speedup. But which costs like crazy from what I hear.
I'm hesitant of putting all my eggs into one big basket.
After talking with a FE, it looks like there are all sorts of goodies in 8.2.1 and up, such that we can setup multiple Virtual Filers (now known as SVMs I think) on the cluster and do transparent volume moves across heads/aggregates, etc. And moving a SVM from one cluster head to another or to another set of nodes in the cluster.
My question of course, is how well dows this work in practise? Esp for FCP luns. Being able to live move volumes between aggregates looks to be a real help, since now I won't be stuck withe large, but not endless aggregates.
Basically, I'm looking for what people think of how well this all works in real life and what the gotchas are. And of course, what about tape backups over direct attached SAN FC tape drives from an SVM? How is that performance as well?
Thanks, John _______________________________________________ Toasters mailing list Toasters@teaparty.net http://www.teaparty.net/mailman/listinfo/toasters
There are limits to the maximum number of nodes in a cluster, which varies depending on the model. If I recall correctly, FAS6200 series with NO iSCSI or FCP can go up to 24 nodes. FAS6200 with iSCSI or FC can go up to 8 nodes. Mixing FAS3200 with FAS6200 means max of 8 nodes.
We're in the process of transitioning some of our NFS customers over to a CMode system. We'll wait a bit for CIFS to be more baked, since only with the RC does it support offboard virus scanning and some other features 7mode has had for ages.
-- Mike Garrison
On Thu, Dec 19, 2013 at 5:14 PM, John Stoffel john@stoffel.org wrote:
Guys,
We're looking to spec out a big Netapp cluster, and since 7-mode is end of life in terms of new features, and it looks like 8.2.1RC1 is now out, what do people think of it?
One question I have is whether I should get a bunch of 3250s in cluster pairs, on the order of four or six heads all clustered together, or whether it makes more sense to get a 6250 pair (or pairs?) instead.
We're looking at roughtly 1 Petabyte of storage. Probably 50% can be SATA, the rest SAS, but possibly <20% would need to be FC LUNs for Oracle DBs doing some ERP stuff. Lots of IOPs will happen there. Another thought is to get SSDs and setup FlexPools. Which is nice since you can prioritze volumes to get that speedup. But which costs like crazy from what I hear.
I'm hesitant of putting all my eggs into one big basket.
After talking with a FE, it looks like there are all sorts of goodies in 8.2.1 and up, such that we can setup multiple Virtual Filers (now known as SVMs I think) on the cluster and do transparent volume moves across heads/aggregates, etc. And moving a SVM from one cluster head to another or to another set of nodes in the cluster.
My question of course, is how well dows this work in practise? Esp for FCP luns. Being able to live move volumes between aggregates looks to be a real help, since now I won't be stuck withe large, but not endless aggregates.
Basically, I'm looking for what people think of how well this all works in real life and what the gotchas are. And of course, what about tape backups over direct attached SAN FC tape drives from an SVM? How is that performance as well?
Thanks, John _______________________________________________ Toasters mailing list Toasters@teaparty.net http://www.teaparty.net/mailman/listinfo/toasters
Lun moves work fine on cluster mode. It also works on 7-mode for aggregates on the same filer.
Cluster mode is managed differently , allow for time for training and lots of testing in your environment.
A drawback is that luns cannot be migrated without a third party appliance and a lengthy maintenance window.
On Dec 19, 2013, at 5:57 PM, Michael Garrison mcgarr@umich.edu wrote:
There are limits to the maximum number of nodes in a cluster, which varies depending on the model. If I recall correctly, FAS6200 series with NO iSCSI or FCP can go up to 24 nodes. FAS6200 with iSCSI or FC can go up to 8 nodes. Mixing FAS3200 with FAS6200 means max of 8 nodes.
We're in the process of transitioning some of our NFS customers over to a CMode system. We'll wait a bit for CIFS to be more baked, since only with the RC does it support offboard virus scanning and some other features 7mode has had for ages. Pun -- Mike Garrison
On Thu, Dec 19, 2013 at 5:14 PM, John Stoffel john@stoffel.org wrote:
Guys,
We're looking to spec out a big Netapp cluster, and since 7-mode is end of life in terms of new features, and it looks like 8.2.1RC1 is now out, what do people think of it?
One question I have is whether I should get a bunch of 3250s in cluster pairs, on the order of four or six heads all clustered together, or whether it makes more sense to get a 6250 pair (or pairs?) instead.
We're looking at roughtly 1 Petabyte of storage. Probably 50% can be SATA, the rest SAS, but possibly <20% would need to be FC LUNs for Oracle DBs doing some ERP stuff. Lots of IOPs will happen there. Another thought is to get SSDs and setup FlexPools. Which is nice since you can prioritze volumes to get that speedup. But which costs like crazy from what I hear.
I'm hesitant of putting all my eggs into one big basket.
After talking with a FE, it looks like there are all sorts of goodies in 8.2.1 and up, such that we can setup multiple Virtual Filers (now known as SVMs I think) on the cluster and do transparent volume moves across heads/aggregates, etc. And moving a SVM from one cluster head to another or to another set of nodes in the cluster.
My question of course, is how well dows this work in practise? Esp for FCP luns. Being able to live move volumes between aggregates looks to be a real help, since now I won't be stuck withe large, but not endless aggregates.
Basically, I'm looking for what people think of how well this all works in real life and what the gotchas are. And of course, what about tape backups over direct attached SAN FC tape drives from an SVM? How is that performance as well?
Thanks, John _______________________________________________ Toasters mailing list Toasters@teaparty.net http://www.teaparty.net/mailman/listinfo/toasters
Toasters mailing list Toasters@teaparty.net http://www.teaparty.net/mailman/listinfo/toasters
This may be a silly question, and is certainly a religious debate, but, ah, any reason you must put those Oracle databases on LUNs? Why not just use NFS?
-Adam
On Thu, Dec 19, 2013 at 6:31 PM, Mark mkopenski@gmail.com wrote:
Lun moves work fine on cluster mode. It also works on 7-mode for aggregates on the same filer.
Cluster mode is managed differently , allow for time for training and lots of testing in your environment.
A drawback is that luns cannot be migrated without a third party appliance and a lengthy maintenance window.
On Dec 19, 2013, at 5:57 PM, Michael Garrison mcgarr@umich.edu wrote:
There are limits to the maximum number of nodes in a cluster, which varies depending on the model. If I recall correctly, FAS6200 series with NO iSCSI or FCP can go up to 24 nodes. FAS6200 with iSCSI or FC can go up to 8 nodes. Mixing FAS3200 with FAS6200 means max of 8 nodes.
We're in the process of transitioning some of our NFS customers over to a CMode system. We'll wait a bit for CIFS to be more baked, since only with the RC does it support offboard virus scanning and some other features 7mode has had for ages. Pun -- Mike Garrison
On Thu, Dec 19, 2013 at 5:14 PM, John Stoffel john@stoffel.org wrote:
Guys,
We're looking to spec out a big Netapp cluster, and since 7-mode is end of life in terms of new features, and it looks like 8.2.1RC1 is now out, what do people think of it?
One question I have is whether I should get a bunch of 3250s in cluster pairs, on the order of four or six heads all clustered together, or whether it makes more sense to get a 6250 pair (or pairs?) instead.
We're looking at roughtly 1 Petabyte of storage. Probably 50% can be SATA, the rest SAS, but possibly <20% would need to be FC LUNs for Oracle DBs doing some ERP stuff. Lots of IOPs will happen there. Another thought is to get SSDs and setup FlexPools. Which is nice since you can prioritze volumes to get that speedup. But which costs like crazy from what I hear.
I'm hesitant of putting all my eggs into one big basket.
After talking with a FE, it looks like there are all sorts of goodies in 8.2.1 and up, such that we can setup multiple Virtual Filers (now known as SVMs I think) on the cluster and do transparent volume moves across heads/aggregates, etc. And moving a SVM from one cluster head to another or to another set of nodes in the cluster.
My question of course, is how well dows this work in practise? Esp for FCP luns. Being able to live move volumes between aggregates looks to be a real help, since now I won't be stuck withe large, but not endless aggregates.
Basically, I'm looking for what people think of how well this all works in real life and what the gotchas are. And of course, what about tape backups over direct attached SAN FC tape drives from an SVM? How is that performance as well?
Thanks, John _______________________________________________ Toasters mailing list Toasters@teaparty.net http://www.teaparty.net/mailman/listinfo/toasters
Toasters mailing list Toasters@teaparty.net http://www.teaparty.net/mailman/listinfo/toasters
Toasters mailing list Toasters@teaparty.net http://www.teaparty.net/mailman/listinfo/toasters
I think it depends on your environment. AFAIK the one thing Fibre still has going for it is that it's *very* deterministic - you know what your latency will be and the level of jitter is very small. That does not mean it's "faster" than IP based connectivity but for some applications I can see *knowing* you are inside a very narrow set of parameters as being worth the extra hassle of maintaining a FC infrastructure. That should be very small percentage of the IT population though. I am quite happy with Oracle's performance over NFS.
On 12/20/2013 08:50 AM, Adam Levin wrote:
This may be a silly question, and is certainly a religious debate, but, ah, any reason you must put those Oracle databases on LUNs? Why not just use NFS?
-Adam
On Thu, Dec 19, 2013 at 6:31 PM, Mark <mkopenski@gmail.com mailto:mkopenski@gmail.com> wrote:
Lun moves work fine on cluster mode. It also works on 7-mode for aggregates on the same filer. Cluster mode is managed differently , allow for time for training and lots of testing in your environment. A drawback is that luns cannot be migrated without a third party appliance and a lengthy maintenance window. On Dec 19, 2013, at 5:57 PM, Michael Garrison <mcgarr@umich.edu <mailto:mcgarr@umich.edu>> wrote:There are limits to the maximum number of nodes in a cluster, which varies depending on the model. If I recall correctly, FAS6200 series with NO iSCSI or FCP can go up to 24 nodes. FAS6200 with iSCSI or FC can go up to 8 nodes. Mixing FAS3200 with FAS6200 means max of 8 nodes. We're in the process of transitioning some of our NFS customers over to a CMode system. We'll wait a bit for CIFS to be more baked, since only with the RC does it support offboard virus scanning and some other features 7mode has had for ages. Pun -- Mike Garrison On Thu, Dec 19, 2013 at 5:14 PM, John Stoffel <john@stoffel.org <mailto:john@stoffel.org>> wrote: Guys, We're looking to spec out a big Netapp cluster, and since 7-mode is end of life in terms of new features, and it looks like 8.2.1RC1 is now out, what do people think of it? One question I have is whether I should get a bunch of 3250s in cluster pairs, on the order of four or six heads all clustered together, or whether it makes more sense to get a 6250 pair (or pairs?) instead. We're looking at roughtly 1 Petabyte of storage. Probably 50% can be SATA, the rest SAS, but possibly <20% would need to be FC LUNs for Oracle DBs doing some ERP stuff. Lots of IOPs will happen there. Another thought is to get SSDs and setup FlexPools. Which is nice since you can prioritze volumes to get that speedup. But which costs like crazy from what I hear. I'm hesitant of putting all my eggs into one big basket. After talking with a FE, it looks like there are all sorts of goodies in 8.2.1 and up, such that we can setup multiple Virtual Filers (now known as SVMs I think) on the cluster and do transparent volume moves across heads/aggregates, etc. And moving a SVM from one cluster head to another or to another set of nodes in the cluster. My question of course, is how well dows this work in practise? Esp for FCP luns. Being able to live move volumes between aggregates looks to be a real help, since now I won't be stuck withe large, but not endless aggregates. Basically, I'm looking for what people think of how well this all works in real life and what the gotchas are. And of course, what about tape backups over direct attached SAN FC tape drives from an SVM? How is that performance as well? Thanks, John _______________________________________________ Toasters mailing list Toasters@teaparty.net <mailto:Toasters@teaparty.net> http://www.teaparty.net/mailman/listinfo/toasters _______________________________________________ Toasters mailing list Toasters@teaparty.net <mailto:Toasters@teaparty.net> http://www.teaparty.net/mailman/listinfo/toasters_______________________________________________ Toasters mailing list Toasters@teaparty.net <mailto:Toasters@teaparty.net> http://www.teaparty.net/mailman/listinfo/toasters
Toasters mailing list Toasters@teaparty.net http://www.teaparty.net/mailman/listinfo/toasters
Please be advised that this email may contain confidential information. If you are not the intended recipient, please notify us by email by replying to the sender and delete this message. The sender disclaims that the content of this email constitutes an offer to enter into, or the acceptance of, any agreement; provided that the foregoing does not invalidate the binding effect of any digital or other electronic reproduction of a manual signature that is included in any attachment.
Try using the ORACLE NFS implimentation... It really speeds things up if you can use multiple interfaces on your oracle server and your NetApp. (i.e. you do not need to use any ether-channel. Oracle's NFS will spread out over all interfaces you tell it about)
--tmac
*Tim McCarthy* *Principal Consultant*
Clustered ONTAP Clustered ONTAP NCDA ID: XK7R3GEKC1QQ2LVD RHCE6 110-107-141https://www.redhat.com/wapps/training/certification/verify.html?certNumber=110-107-141&isSearch=False&verify=Verify NCSIE ID: C14QPHE21FR4YWD4 Expires: 08 November 2014 Current until Aug 02, 2016 Expires: 08 November 2014
On Fri, Dec 20, 2013 at 8:58 AM, Jeremy Page jeremy.page@gilbarco.comwrote:
I think it depends on your environment. AFAIK the one thing Fibre still has going for it is that it's *very* deterministic - you know what your latency will be and the level of jitter is very small. That does not mean it's "faster" than IP based connectivity but for some applications I can see *knowing* you are inside a very narrow set of parameters as being worth the extra hassle of maintaining a FC infrastructure. That should be very small percentage of the IT population though. I am quite happy with Oracle's performance over NFS.
On 12/20/2013 08:50 AM, Adam Levin wrote:
This may be a silly question, and is certainly a religious debate, but, ah, any reason you must put those Oracle databases on LUNs? Why not just use NFS?
-Adam
On Thu, Dec 19, 2013 at 6:31 PM, Mark mkopenski@gmail.com wrote:
Lun moves work fine on cluster mode. It also works on 7-mode for aggregates on the same filer.
Cluster mode is managed differently , allow for time for training and lots of testing in your environment.
A drawback is that luns cannot be migrated without a third party appliance and a lengthy maintenance window.
On Dec 19, 2013, at 5:57 PM, Michael Garrison mcgarr@umich.edu wrote:
There are limits to the maximum number of nodes in a cluster, which varies depending on the model. If I recall correctly, FAS6200 series with NO iSCSI or FCP can go up to 24 nodes. FAS6200 with iSCSI or FC can go up to 8 nodes. Mixing FAS3200 with FAS6200 means max of 8 nodes.
We're in the process of transitioning some of our NFS customers over to a CMode system. We'll wait a bit for CIFS to be more baked, since only with the RC does it support offboard virus scanning and some other features 7mode has had for ages. Pun -- Mike Garrison
On Thu, Dec 19, 2013 at 5:14 PM, John Stoffel john@stoffel.org wrote:
Guys,
We're looking to spec out a big Netapp cluster, and since 7-mode is end of life in terms of new features, and it looks like 8.2.1RC1 is now out, what do people think of it?
One question I have is whether I should get a bunch of 3250s in cluster pairs, on the order of four or six heads all clustered together, or whether it makes more sense to get a 6250 pair (or pairs?) instead.
We're looking at roughtly 1 Petabyte of storage. Probably 50% can be SATA, the rest SAS, but possibly <20% would need to be FC LUNs for Oracle DBs doing some ERP stuff. Lots of IOPs will happen there. Another thought is to get SSDs and setup FlexPools. Which is nice since you can prioritze volumes to get that speedup. But which costs like crazy from what I hear.
I'm hesitant of putting all my eggs into one big basket.
After talking with a FE, it looks like there are all sorts of goodies in 8.2.1 and up, such that we can setup multiple Virtual Filers (now known as SVMs I think) on the cluster and do transparent volume moves across heads/aggregates, etc. And moving a SVM from one cluster head to another or to another set of nodes in the cluster.
My question of course, is how well dows this work in practise? Esp for FCP luns. Being able to live move volumes between aggregates looks to be a real help, since now I won't be stuck withe large, but not endless aggregates.
Basically, I'm looking for what people think of how well this all works in real life and what the gotchas are. And of course, what about tape backups over direct attached SAN FC tape drives from an SVM? How is that performance as well?
Thanks, John _______________________________________________ Toasters mailing list Toasters@teaparty.net http://www.teaparty.net/mailman/listinfo/toasters
_______________________________________________Toasters mailing list Toasters@teaparty.net http://www.teaparty.net/mailman/listinfo/toasters
Toasters mailing list Toasters@teaparty.net http://www.teaparty.net/mailman/listinfo/toasters
Toasters mailing listToasters@teaparty.nethttp://www.teaparty.net/mailman/listinfo/toasters
Please be advised that this email may contain confidential information. If you are not the intended recipient, please notify us by email by replying to the sender and delete this message. The sender disclaims that the content of this email constitutes an offer to enter into, or the acceptance of, any agreement; provided that the foregoing does not invalidate the binding effect of any digital or other electronic reproduction of a manual signature that is included in any attachment.
Toasters mailing list Toasters@teaparty.net http://www.teaparty.net/mailman/listinfo/toasters
tmac> Try using the ORACLE NFS implimentation...
Oracle 11g I assume?
tmac> It really speeds things up if you can use multiple interfaces on your tmac> oracle server and your NetApp.
How about a single 10G pair?
tmac> (i.e. you do not need to use any ether-channel. Oracle's NFS tmac> will spread out over all interfaces you tell it about)
Nice to know. I really would like to make Oracle over NFS work, but it might be an uphill battle to convince people.
John
It is not the same. E.g. newer Linux kernel will use single TCP connection to NFS server even if you have multiple mount points. This did prove to become a bottleneck, even with 10G.
Отправлено с iPhone
20 дек. 2013 г., в 19:49, "John Stoffel" john@stoffel.org написал(а):
tmac> Try using the ORACLE NFS implimentation...
Oracle 11g I assume?
tmac> It really speeds things up if you can use multiple interfaces on your tmac> oracle server and your NetApp.
How about a single 10G pair?
I was wondering if I can install OnCommand Core 5.2 and during the install, use an existing 4.02 db backup on Linux x64? I was under the impression I would have to install DMF 4.02, import the db during the setup, then do an upgrade to OCC5.2... I know there is a big difference between 5.1 and 5.2 to address the db delete issues.. Just wondering if anyone has done this before.
happy new year everyone!!
Steve
Adam> This may be a silly question, and is certainly a religious Adam> debate, but, ah, any reason you must put those Oracle databases Adam> on LUNs? Why not just use NFS?
It's the religon of the Oracle Apps people, and the lack of really good high speed network at the time. I personally think that putting Development Oracle on NFS would be a win win, but I'm not able to win that battle.
And we do have teams with Oracle on NFS over 10G ethernet, so it's possible to do.
My current 7-mode cluster has Oracle on SAN and its painful at times in terms of managing, and the stupidity of the Oracle admins to not re-write queries to be faster or more efficient. But that's religious reasons.
I have to accept reality and just make it work, and SAN FCP Luns are known to work. So if I can do it, great. If not... then I need to do something else.
Which is all a long roundabout way to say I agree with you. :-)
John
Mark> Lun moves work fine on cluster mode. It also works on 7-mode for Mark> aggregates on the same filer.
Good to know. Are there any gotchas to keep an eye out for?
Mark> Cluster mode is managed differently , allow for time for Mark> training and lots of testing in your environment.
Yeah, I need to download and install the Netapp ESX images and start playing....
Mark> A drawback is that luns cannot be migrated without a third party Mark> appliance and a lengthy maintenance window.
Umm... you've confused me here. What is the difference between move and migrate? Is Migrate moving the LUN to a differnet head, while move is just moving between aggregates on the same head?
And what third party appliance would you be talking about here? I think alot of people would be interested in knowing more details here, since it's all still pretty new.
John
What I meant that cannot be migrated from a 7-mode to a cluster mode tool without an external appliance called the dta
On Dec 20, 2013, at 10:08 AM, "John Stoffel" john@stoffel.org wrote:
Mark> Lun moves work fine on cluster mode. It also works on 7-mode for Mark> aggregates on the same filer.
Good to know. Are there any gotchas to keep an eye out for?
Mark> Cluster mode is managed differently , allow for time for Mark> training and lots of testing in your environment.
Yeah, I need to download and install the Netapp ESX images and start playing....
Mark> A drawback is that luns cannot be migrated without a third party Mark> appliance and a lengthy maintenance window.
Umm... you've confused me here. What is the difference between move and migrate? Is Migrate moving the LUN to a differnet head, while move is just moving between aggregates on the same head?
And what third party appliance would you be talking about here? I think alot of people would be interested in knowing more details here, since it's all still pretty new.
John
Mark> What I meant that cannot be migrated from a 7-mode to a cluster Mark> mode tool without an external appliance called the dta
Ah, now I understand. and it makes perfect sense. In this situation, we'd probably just snapmirror the data to the new cluster and start from semi-scratch.
John
That is what Mark is referring to, unless your comment refers to migrating from one cluster to another. You can not SnapMirror from 7-mode to cluster-mode (clustered Data ONTAP aka cDOT). There is a tool available on the support site called the 7-mode transition tool (7MTT). This does leverage SnapMirror between the two versions but only for the purpose of migration and currently only for file data (CIFS,NFS) not block (FC,iSCSI,etc). There is an appliance used to move block data called a DTA.
Daniel DeCristofaro Systems Engineer
NetApp 585.789.1007 Mobile Phone daniel.decristofaro@netapp.com mailto:name@netapp.com www.netapp.com/us/technology/agile-data-infrastructure.html http://www.netapp.com/us/technology/agile-data-infrastructure.html http://www.netapp.com/us/technology/agile-data-infrastructure.html
Follow us: http://www.facebook.com/NetApp?REF_SOURCE=ems-facebook http://twitter.com/#netapp?REF_SOURCE=ems-twitter <http://www.linkedin.com/groups/NetApp-111681/about?REF_SOURCE=ems-linkedin
http://www.youtube.com/user/NetAppTV?REF_SOURCE=ems-youtube http://www.slideshare.net/NetApp?REF_SOURCE=ems-slideshare https://communities.netapp.com/welcome?REF_SOURCE=ems-ctyTweet us:#netapp
On 12/20/13 2:59 PM, "John Stoffel" john@stoffel.org wrote:
Mark> What I meant that cannot be migrated from a 7-mode to a cluster Mark> mode tool without an external appliance called the dta
Ah, now I understand. and it makes perfect sense. In this situation, we'd probably just snapmirror the data to the new cluster and start from semi-scratch.
John
Toasters mailing list Toasters@teaparty.net http://www.teaparty.net/mailman/listinfo/toasters
Snapmirror migration for luns from 7-mode to cDOT will be available in ONTAP 8.3 as far as I've heard. That would be about 6 months away.
/ Marcus
-----Original Message----- From: toasters-bounces@teaparty.net [mailto:toasters- bounces@teaparty.net] On Behalf Of DeCristofaro, Daniel Sent: den 20 december 2013 21:25 To: John Stoffel; Mark Kopenski Cc: toasters@teaparty.net Subject: Re: OnTap cluster mode questions
That is what Mark is referring to, unless your comment refers to migrating from one cluster to another. You can not SnapMirror from 7-mode to cluster-mode (clustered Data ONTAP aka cDOT). There is a tool available
on
the support site called the 7-mode transition tool (7MTT). This does
leverage
SnapMirror between the two versions but only for the purpose of migration and currently only for file data (CIFS,NFS) not block (FC,iSCSI,etc).
There is an
appliance used to move block data called a DTA.
Daniel DeCristofaro Systems Engineer
NetApp 585.789.1007 Mobile Phone daniel.decristofaro@netapp.com mailto:name@netapp.com www.netapp.com/us/technology/agile-data-infrastructure.html http://www.netapp.com/us/technology/agile-data-infrastructure.html http://www.netapp.com/us/technology/agile-data-infrastructure.html
Follow us: http://www.facebook.com/NetApp?REF_SOURCE=ems- facebook http://twitter.com/#netapp?REF_SOURCE=ems-twitter <http://www.linkedin.com/groups/NetApp- 111681/about?REF_SOURCE=ems-linkedin
http://www.youtube.com/user/NetAppTV?REF_SOURCE=ems-youtube http://www.slideshare.net/NetApp?REF_SOURCE=ems-slideshare <https://communities.netapp.com/welcome?REF_SOURCE=ems-
cty>Tweet
us:#netapp
On 12/20/13 2:59 PM, "John Stoffel" john@stoffel.org wrote:
Mark> What I meant that cannot be migrated from a 7-mode to a cluster Mark> mode tool without an external appliance called the dta
Ah, now I understand. and it makes perfect sense. In this situation, we'd probably just snapmirror the data to the new cluster and start from semi-scratch.
John
Toasters mailing list Toasters@teaparty.net http://www.teaparty.net/mailman/listinfo/toasters
Toasters mailing list Toasters@teaparty.net http://www.teaparty.net/mailman/listinfo/toasters
We are in the process of migrating our 7-mode systems to a cDOT cluster running cDOT 8.2P3 (we will be updating again soon). The 7MTT tool works well however I have noticed it tends to leave some cruft behind. I have created a volume migration checklist that I use for every NAS volume migration that I have performed. If you would like a copy feel free to email me directly and I can provide that to you.
As far as LUN-based volumes go we opted to use client-based utilities such as robocopy to migrate the data to the new LUNs. This option isn't for everybody however we are primarily a NAS shop so there is not much LUN-based data to worry about.
Dan
Sent from my mobile device, please excuse typos.
On Dec 20, 2013, at 2:27 PM, "DeCristofaro, Daniel" Daniel.DeCristofaro@netapp.com wrote:
That is what Mark is referring to, unless your comment refers to migrating from one cluster to another. You can not SnapMirror from 7-mode to cluster-mode (clustered Data ONTAP aka cDOT). There is a tool available on the support site called the 7-mode transition tool (7MTT). This does leverage SnapMirror between the two versions but only for the purpose of migration and currently only for file data (CIFS,NFS) not block (FC,iSCSI,etc). There is an appliance used to move block data called a DTA.
Daniel DeCristofaro Systems Engineer
NetApp 585.789.1007 Mobile Phone daniel.decristofaro@netapp.com mailto:name@netapp.com www.netapp.com/us/technology/agile-data-infrastructure.html http://www.netapp.com/us/technology/agile-data-infrastructure.html http://www.netapp.com/us/technology/agile-data-infrastructure.html
Follow us: http://www.facebook.com/NetApp?REF_SOURCE=ems-facebook http://twitter.com/#netapp?REF_SOURCE=ems-twitter <http://www.linkedin.com/groups/NetApp-111681/about?REF_SOURCE=ems-linkedin
http://www.youtube.com/user/NetAppTV?REF_SOURCE=ems-youtube http://www.slideshare.net/NetApp?REF_SOURCE=ems-slideshare https://communities.netapp.com/welcome?REF_SOURCE=ems-ctyTweet us:#netapp
On 12/20/13 2:59 PM, "John Stoffel" john@stoffel.org wrote:
Mark> What I meant that cannot be migrated from a 7-mode to a cluster Mark> mode tool without an external appliance called the dta
Ah, now I understand. and it makes perfect sense. In this situation, we'd probably just snapmirror the data to the new cluster and start from semi-scratch.
John
Toasters mailing list Toasters@teaparty.net http://www.teaparty.net/mailman/listinfo/toasters
Toasters mailing list Toasters@teaparty.net http://www.teaparty.net/mailman/listinfo/toasters
Michael> There are limits to the maximum number of nodes in a cluster, which varies Michael> depending on the model. If I recall correctly, FAS6200 series with NO iSCSI Michael> or FCP can go up to 24 nodes. FAS6200 with iSCSI or FC can go up to 8 Michael> nodes. Mixing FAS3200 with FAS6200 means max of 8 nodes.
Hmm... as I understand it, the maximun number of nodes in a FCP cluster is 8, due to ALUA issues. I didn't get a good answer about whether you could have a subset of nodes doing FCP, with the rest of the nodes doing NFS/CIFS.
Michael> We're in the process of transitioning some of our NFS Michael> customers over to a CMode system. We'll wait a bit for CIFS Michael> to be more baked, since only with the RC does it support Michael> offboard virus scanning and some other features 7mode has had Michael> for ages.
That's a good point to keep in mind.
Thanks for the feedback.
John
-
Adam Levin -
Borzenkov, Andrey -
Dan Burkland -
DeCristofaro, Daniel -
Jeremy Page -
John Stoffel -
Marcus Nilsson -
Mark -
Mark Kopenski
-
Michael Garrison -
steve klise -
tmac