You are viewing a plain text version of this content. The canonical link for it is here.
Posted to users@cloudstack.apache.org by Go Chiba <go...@gmail.com> on 2013/03/04 17:31:12 UTC
Question for storage traffic
Hi folks,
I have a question for traffic type that discussed as CLOUDSTACK-382 on JIRA.
ref:
https://issues.apache.org/jira/browse/CLOUDSTACK-382?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
By my understanding, "storage traffic" isolation on CS means separate disk
I/O for primary storage as that explained blog post by shapeblue team and I
believe their explanation is correct.
But could you clearly my understanding for architecture of CS networking.
Because, official document cause misunderstanding or confusion as me and it
caused business impact for design of production services of several
enterprise users...
regards
--
Go Chiba
E-mail:go.chiba@gmail.com
RE: Question for storage traffic
Posted by Geoff Higginbottom <ge...@shapeblue.com>.
Hi Chip,
You are absolutely right, configuring a Management Interface on XenServer, using a Primary Storage CIDR, which is different to the main Management Interface CIDR, will send traffic destined for Storage over that interface (Bond).
We typically create the Storage Bond with XenCenter before adding the Host to CloudStack, then after adding the Host, add Primary Storage using the CloudStack API specifying the Primary Storage IP or FQDN and Cloudstack takes care of the rest.
Regards
Geoff Higginbottom
D: +44 20 3603 0542 | S: +44 20 3603 0540 | M: +447968161581
geoff.higginbottom@shapeblue.com
-----Original Message-----
From: Chip Childers [mailto:chip.childers@sungard.com]
Sent: 04 March 2013 18:38
To: cloudstack-users@incubator.apache.org
Subject: Re: Question for storage traffic
On Tue, Mar 05, 2013 at 03:21:08AM +0900, Go Chiba wrote:
> Chip,
>
> Is it means current CS not supported isolation for traffic between
> hypervisor and primary storage?
> And it's only supported secondary storage traffic on CS natively.
> So, if we require design isolation of VM disk I/O out of CS architecture.
>
> Is it right?
>
> Go Chiba
Someone feel free to correct me if there is a newer answer...
But yes, if you want to send traffic from the host to primary storage via a dedicated NIC, I've found that you have to implement the datastore on the host before telling CloudStack about it.
ShapeBlue provides a range of strategic and technical consulting and implementation services to help IT Service Providers and Enterprises to build a true IaaS compute cloud. ShapeBlue’s expertise, combined with CloudStack technology, allows IT Service Providers and Enterprises to deliver true, utility based, IaaS to the customer or end-user.
________________________________
This email and any attachments to it may be confidential and are intended solely for the use of the individual to whom it is addressed. Any views or opinions expressed are solely those of the author and do not necessarily represent those of Shape Blue Ltd. If you are not the intended recipient of this email, you must neither take any action based upon its contents, nor copy or show it to anyone. Please contact the sender if you believe you have received this email in error. Shape Blue Ltd is a company incorporated in England & Wales.
RE: Question for storage traffic
Posted by Clayton Weise <cw...@iswest.net>.
> But yes, if you want to send traffic from the host to primary storage
> via a dedicated NIC, I've found that you have to implement the datastore
> on the host before telling CloudStack about it.
For the most part, yes. While in some cases you don't have to implement the datastore you have to at least get the network configured properly. For example, if you had an iSCSI datastore that you wanted to use, you can tell CS to have your host mount it but beforehand you must configure that iSCSI interface (or interfaces if bonding) on your host before having CS configure the datastore. Generally speaking, even though this is possible I've found configuring it first and telling CS that it's 'preSetup' is less prone to issues.
With KVM and sharedMountPoint you _must_ configure the storage beforehand since it has to exist. With NFS on any system you don't have to mount the NFS export ahead of time, you just need to make sure the host _can_ mount the NFS export and CS will take care of the rest.
Re: Question for storage traffic
Posted by Chip Childers <ch...@sungard.com>.
On Tue, Mar 05, 2013 at 03:21:08AM +0900, Go Chiba wrote:
> Chip,
>
> Is it means current CS not supported isolation for traffic between
> hypervisor and primary storage?
> And it's only supported secondary storage traffic on CS natively.
> So, if we require design isolation of VM disk I/O out of CS architecture.
>
> Is it right?
>
> Go Chiba
Someone feel free to correct me if there is a newer answer...
But yes, if you want to send traffic from the host to primary storage
via a dedicated NIC, I've found that you have to implement the datastore
on the host before telling CloudStack about it.
Re: Question for storage traffic
Posted by Go Chiba <go...@gmail.com>.
Chip,
Is it means current CS not supported isolation for traffic between
hypervisor and primary storage?
And it's only supported secondary storage traffic on CS natively.
So, if we require design isolation of VM disk I/O out of CS architecture.
Is it right?
Go Chiba
Re: Question for storage traffic
Posted by Chip Childers <ch...@sungard.com>.
On Tue, Mar 05, 2013 at 01:31:12AM +0900, Go Chiba wrote:
> Hi folks,
>
> I have a question for traffic type that discussed as CLOUDSTACK-382 on JIRA.
> ref:
> https://issues.apache.org/jira/browse/CLOUDSTACK-382?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
>
> By my understanding, "storage traffic" isolation on CS means separate disk
> I/O for primary storage as that explained blog post by shapeblue team and I
> believe their explanation is correct.
> But could you clearly my understanding for architecture of CS networking.
> Because, official document cause misunderstanding or confusion as me and it
> caused business impact for design of production services of several
> enterprise users...
>
> regards
> --
> Go Chiba
> E-mail:go.chiba@gmail.com
When the documentation says "storage traffic", it's actually talking
about traffic to and from the secondary storage of the zone. The point
of that bug, is that it's obviously quite confusing to simply call it
"storage traffic". What we've done, in order to isolate traffic
intended for *primary storage* (Xen environment at least) to a dedicated
pair of NICs on the hosts, is to pre-create the storage volumes in the
Xen clusters. Then, CS simply knows about the pre-existing primary storage,
and you are able to have at least host to primary storage traffic
running through it's own cables to the top of rack switches.
-chip