Re: [sami] Bringing new work into the IETF

["So, Ning" <[e-mail]>]

David,

I agree with you that the problems are not clearly defined, the requirements are not clear, and the scope is too large.  However, I disagree that providers do not want standardized solution. At least that is not the case for the traditional telecom service providers who are also trying to become the cloud/data center service providers. 

Telecom service provider such as Verizon have significant business with enterprises and government agencies.  One significant cloud service business for that market in the foreseeable future is to host overflow/expansion capacity for the enterprise/government data centers.  That means the provider data centers will have to be multi-tenant in nature with seamless inter-working with the customer data centers.  There are quite a few (I cannot number them) hypervisors out there in the market today, and the customers I know have all of them (although usually each customer has one or two hypervisors only).   Without standardization means the provider data center has to have all of the hypervisors in each and every data center in order to provide the services, and it also means the server/network capacity in the provider data centers are dedicated per Hypervisor without capacity sharing.  

Of course I am only stating a large problem for some providers, not the ones like GOOGLE and YAHOO who have total control of their data centers so they can deploy the same hypervisor everywhere.  And I also understand vendors love the proprietary environment because wasted provider capacity means more business for them.  However, providers are not all dummies.  We will eventually see what is happening, and we will use the well-established equipment selection process (RFP) to drive standardization in order to become more efficient.  

Verizon has always been big on networking technology standardization, and I don't see any difference when it comes to data center and cloud.   From my perspective, standardization is good for IETF, and good for the industry as a whole.  
 
 
Best regards,
 
Ning So
Verizon Corporate Technology
(office) 972-729-7905
(Cell) 972-955-0914
 

-----Original Message-----
From: [e-mail] [mailto:[e-mail]] On Behalf Of David Harrington
Sent: Thursday, August 18, 2011 10:25 AM
To: 'Yingjie Gu(yingjie)'; [e-mail]
Subject: [sami] Bringing new work into the IETF

Hi Yingjie,

I trimmed the CC: list to the sami mailing list.
Those who are interested in the sami discussion can join the sami list and find this discussion in the sami archive.

So far, I think your proposal is not being accepted by the IETF because it is too general and too abstract.
You are still trying to figure out what problem you want to solve.

To make VM migration standardization a viable problem to work on, you probably need to get vendors and operators to all agree on complete standardization for their various technologies - virtualization, TCP/IP stacks, network management protocols, and so on. 
I think you are very unlikely to ever achieve that because vendors want to continue to enhance their solutions versus their competitors.
Standardization is not appropriate to solve every problem. 

Most enterprises/providers will choose a specific virtualization solution, such as VMware.
How VM migration is done depends a great deal on the virtualization solution in use.
Hypervisors have not been standardized because the companies that provide virtualization solutions gain competitive advantage from their hypervisor capabilities. They don't necessarily want to standardize their hypervisors across vendors.

Most enterprises/providers will choose a specific storage solution, such as EMC or NetApp.
How VM migration is done depends a great deal on the storage solution in use.
Storage devices usually use standards, but have not been completely standardized because the companies that provide storage solutions gain competitive advantage from their storage solution enhancements. They don't want to totally standardize storage across vendors.

Most enterprises/providers will choose specific equipment vendors to deploy in their networks.
Those choices often have to do with the proprietary enhancements offered by different vendors.
The vendors have no desire to all offer identical products and identical features.
How VM migration is done depends a great deal on the networking equipment in use.
IP/TCP stacks have not been totally standardized because the companies that provide networking equipment gain competitive advantage from their equipment's enhanced capabilities. They don't want to totally standardize their equipment across vendors.

Most enterprises/providers will choose specific equipment to deploy in their networks.
Those choices often have to do with the proprietary solutions to manage the chosen equipment.
The vendors have no desire to all offer identical management features.
How a migration is done depends a great deal on the networking management tools in use.
CLIs and proprietary MIB modules and other management interfaces have not been totally standardized because the companies that provide these tools gain competitive advantage from their management products'
enhanced capabilities. They don't want to totally standardize their management interfaces.

To standardize VM migration across all the VM implementations, the TCP/IP implementations, the storage implementations, the routing implementations, the security implementations, the application implementations, the network management implementations, and the enterprise-specific deployment choices, is like "boiling the ocean".
Not a very achieveable goal.

The IETF does **engineering**.
You would do much better to focus your energy on a **small** problem, solvable by standardization, that can be turned into an engineering project appropriate to the IETF. 

For example, you might focus on developing a standard DHCP attribute or a standard netconf/YANG module for some aspect of virtual host configuration that could be reasonably standardized across virtualization platforms and equipment vendors. Each standardization of a small piece of Internet-related VM configuration could help standardize VM migration in the long run. And that would be engineering work appropriate to the IETF. 

David Harrington
Director, IETF Transport Area
[e-mail] (preferred for ietf) [e-mail]
[phone] (cell)
 

> -----Original Message-----
> From: Yingjie Gu(yingjie) [mailto:[e-mail]]
> Sent: Thursday, August 18, 2011 8:03 AM
> To: 'So, Ning'; 'Linda Dunbar'; 'Juergen Schoenwaelder'
> Cc: 'Wesley Eddy'; 'Romascanu, Dan (Dan)'; [e-mail]; 'David 
> Harrington'; [e-mail]
> Subject: Re: [sami] Welcome to SAMI and something you may like to 
> know.
> 
> Ning,
> 
> When we talked at Quebec, I remembered that you have impressive 
> understanding on VM migration requirements in/between subnets. Hope 
> you can share your use cases from provider's perspective.
> I have talked with two providers in detail, and they both see the 
> strong requirements for state migration as VM migrates.
> Perhaps we need to document provider's requirements for people's 
> reference. I am working with providers on this document to introduce 
> use cases.
> 
> 
> 
> Best Regards
> Gu Yingjie
> 
> 
> -----邮件原件-----
> 发件人: So, Ning [mailto:[e-mail]]
> 发送时间: 2011年8月17日 乐乐21:59
> 收件人: Yingjie Gu(yingjie); 'Linda Dunbar'; 'Juergen Schoenwaelder'
> 抄送: 'Wesley Eddy'; 'Romascanu, Dan (Dan)'; [e-mail]; 'David 
> Harrington'; [e-mail]
> 主题: RE: [sami] Welcome to SAMI and something you may like to know.
> 
> Yingjie,
> 
> You might want to consider limiting the scope to problem definition 
> and provider requirements collection first.  I know a lot of work has 
> gone into documenting what is available today.  However, the provider 
> requirements in this
> area is still rather thin.   For example, the requirements 
> for optimization-based migration can be quite different from failure 
> restoration-based migration.  The type of services the VMs are 
> supporting can also impact the migration requirements, thus impacting 
> the solutions.  Defining different types of VM migration and the 
> associated requirements should be a higher priority, in my opinion.
> 
>  
> Best regards,
>  
> Ning So
> Verizon Corporate Technology
> (office) 972-729-7905
> (Cell) 972-955-0914
>  
> 
> -----Original Message-----
> From: [e-mail] [mailto:[e-mail]] On Behalf 
> Of Yingjie Gu(yingjie)
> Sent: Wednesday, August 17, 2011 3:59 AM
> To: 'Linda Dunbar'; 'Juergen Schoenwaelder'
> Cc: 'Wesley Eddy'; 'Romascanu, Dan (Dan)'; [e-mail]; 'David 
> Harrington'; [e-mail]
> Subject: Re: [sami] Welcome to SAMI and something you may like to 
> know.
> 
> The following is a scope suggested by Dr. Fan.
> 
> " According to the current implementation,VM migration is scoped in a 
> layer
> 2 network with shared storage.Key factors that decide whether hot 
> migration is successful includes,from the network's side, bandwith and 
> delay.Even the migration happens between two sites,it may succeed if 
> the bandwidth is wide enough and the delay is small enough.
>  
> So,pehhaps we can define the scope as such:A layer 2 subnet with good 
> enough performance which makes the hot migration successful."
> 
> What is your opinion on this scope?
> 
> Best Regards
> Gu Yingjie
> 
> -----邮件原件-----
> 发件人: Linda Dunbar [mailto:[e-mail]]
> 发送时间: 2011年8月12日 乐乐0:14
> 收件人: Yingjie Gu(yingjie); 'Juergen Schoenwaelder'
> 抄送: 'Wesley Eddy'; 'Romascanu, Dan (Dan)'; [e-mail]; 'David 
> Harrington'; [e-mail]
> 主题: RE: [sami] Welcome to SAMI and something you may like to know.
> 
> I thought at the barBOF that the next step is to identify some use 
> cases. It will be very beneficial to accelerate the progress by 
> identifying some states (or conditions) which today's firewall or 
> security devices can use to continue the proper function.
> 
> 
> Linda
> 
> > -----Original Message-----
> > From: [e-mail] [mailto:[e-mail]]
> On Behalf
> > Of Yingjie Gu(yingjie)
> > Sent: Thursday, August 11, 2011 4:07 AM
> > To: 'Juergen Schoenwaelder'
> > Cc: 'Wesley Eddy'; 'Romascanu, Dan (Dan)'; [e-mail]; 'David 
> > Harrington'; [e-mail]
> > Subject: Re: [sami] Welcome to SAMI and something you may
> like to know.
> > 
> > Hi Juergen,
> > 
> > I accept your suggestion. I will be more careful with my words.
> > 
> > We don't need yet more high level discussion on whether
> it's useful or
> > not.
> > 
> > Let's first figure out the context: scope, use cases and so on.
> > 
> > 
> > Best Regards
> > Gu Yingjie
> > 
> > -----邮件原件-----
> > 发件人: [e-mail] [mailto:[e-mail]] 代表
> > Juergen
> > Schoenwaelder
> > 发送时间: 2011年8月11日 乐乐15:41
> > 收件人: Yingjie Gu(yingjie)
> > 抄送: 'Wesley Eddy'; 'Romascanu, Dan (Dan)';
> [e-mail]; 'David
> > Harrington'; [e-mail]
> > 主题: Re: [sami] Welcome to SAMI and something you may like to
know.
> > 
> > On Thu, Aug 11, 2011 at 02:04:14PM +0800, Yingjie Gu(yingjie)
wrote:
> > 
> > > Attendees agree that state migration is useful. What we need to
do
> > next is
> > > to narrow down the scope, e.g. state migration within the same 
> > > Administration domain or between domains, and collect the
> use cases
> > in
> > that
> > > scope. Then we need to figure out what kind of state need to be
> > migrated
> > in
> > > the narrow scope, what's the common representation of the
states, 
> > > and,
> > if
> > we
> > > go that far, the potential solutions for state migration.
> > 
> > For me, it is crucial to first define the scope before I can agree

> > whether state migration is a useful concept or not. 
> Depending on the
> > timing and the addressing/routing issues, other mechanisms might
be 
> > more appropriate.
> > 
> > Bottom line: Be careful with general statements like
> "Attendees agree
> > that state migration is useful." before we even agree on
> the context.
> > 
> > /js
> > 
> > --
> > Juergen Schoenwaelder           Jacobs University Bremen gGmbH
> > Phone: [phone]         Campus Ring 1, 28759 Bremen,
Germany
> > Fax:   [phone]         <http://www.jacobs-university.de/>
> > _______________________________________________
> > sami mailing list
> > [e-mail]
> > https://www.ietf.org/mailman/listinfo/sami
> > 
> > _______________________________________________
> > sami mailing list
> > [e-mail]
> > https://www.ietf.org/mailman/listinfo/sami
> 
> _______________________________________________
> sami mailing list
> [e-mail]
> https://www.ietf.org/mailman/listinfo/sami
> 
> 

_______________________________________________
sami mailing list
[e-mail]
https://www.ietf.org/mailman/listinfo/sami