Re: [clouds] __答复:_scope_of_the_cloud

Thanks Sam for the reply.
Personally I think the migration,resource description(maybe there are similar efforts in other SDOs already) are good topics to discuss.
As for lightweight HTTP, I am not sure if this is an unique problem in cloud applications.It should be treated as an independent problem in my point of view.
Just two cents.

Yunfei

zhangyunfei
2010-04-09

发件人： Sam Johnston
发送时间： 2010-04-09 17:04:20
收件人： zhangyunfei
抄送： Wang Xu; meng.yu; clouds@ietf.org
主题： Re: [clouds] __答复:_scope_of_the_cloud

Yunfei/Wang/Meng,

My interest was not limited to virtual machines (which require some amount of choreography on either side of the actual move) but rather files, databases, code, documents - basically anything that you might want to move from A to B (where you are not necessarily A or B).

Another area of potential interest is a lightweight "profile" of HTTP for constrained devices, whereby one does not have to implement a full-blown HTTP client & server pair simply to query an electricity meter. Some of these issues were already discussed earlier this week[1]. The same could be said for application of [vanilla] HTTP to files, databases, etc. (e.g. the relevant subset of CDMI).

We're also missing a way of describing containers (e.g. hypervisors, platforms, etc.) and workloads (e.g. applications, virtual machines). CloudAudit provides a technical solution to the former (but it's out of scope) and Cisco's James Urquhart has been working on a descriptor format for the latter[2]. Both of these areas seem appropriate for IETF involvement, and indeed CloudAudit could be worked into a generic catalog service (e.g. the legal name of the provider, its location, logo, services, SLAs, etc.).

I think the best next step would be for to work torwards formalising the "clouds" effort somewhat, which will necessarily involve enumerating the opportunities for standards development and prioritising them. IETF-78 would presumably be a good forum for this activity (assuming it can't be done more inclusively on this mailing list).

Sam

1. http://nih.blogspot.com/2010_04_01_archive.html
2. http://news.cnet.com/8301-19413_3-10440799-240.html

On 9 April 2010 10:07, zhangyunfei <zhangyunfei@chinamobile.com> wrote:

I just want to know more praticial deployment on this point since Sam mentioned this.
----------------------------------
[Sam]
Another issue I ran into while writing cush[3] was how to remotely instruct servers to migrate (live?) resources - for example, moving a virtual machine, database, etc. from one location to another using a mobile device on a 3G connection. I believe WebDAV's COPY and MOVE verbs are a good start (this is what we're using for OCCI) but they could be reviewed and possibly promoted for more generic application.

BR
Yunfei

zhangyunfei
2010-04-09

发件人： Wang Xu
发送时间： 2010-04-09 16:03:32
收件人： meng.yu
抄送： zhangyunfei; clouds@ietf.org
主题： Re:_[clouds]_答复:_scope_of_the_cloud

Yunfei  and  Meng,

Do  you  have  any  practical  use  case  of  live  migration  of  virtual
machine?  I  do  not  find  any  case  of  heavy  utilization  of VM  migration.

And  some  tests  indicated  that  the  migration  in  product  environment  is
not  mature  enough  yet.

2010/4/9   <meng.yu@zte.com.cn >
>
>  Not  only  (virtual  )resource/(or  tasks)  migration  esp.  in  different  clouds,  but  migration  from  different  servers  in  one  cloud,  I  think  it  could  be  resonable.
>
>  Further,  I  am  thinking  could  them  be  harmonized  together  for  the  protocol  aspect?
>
>
>
>  "zhangyunfei"   <zhangyunfei@chinamobile.com >
>  发件人:   clouds-bounces@ietf.org
>
>  2010-04-09  14:21
>
>  收件人
>  "clouds@ietf.org"   <clouds@ietf.org >,  "sjj"   <sjj@google.com >
>  抄送
>  主题
>  [clouds]  scope  of  the  cloud
>
>
>
>
>  Hi  all.
>        Some  kindly  told  me  that  my  email  is  not  displayed  corretly.So  send  again.
>  -------------------------------------------------
>
>  Sam  and  all,
>      The  topic  of  standardizing  (virtual  )resource  migration  esp.  in  different  clouds  you  mentioned  sounds  interesting.Do  you  have  more  details  to  share?Thanks.
>        Another  point:Is  there  standard  protocols  in  interconnecting  different  regions  of  clusters?I  thinks  if  there  are  room  to  expolore,it  would  be  a  suitable  topic  in  IETF.
>
>  BR
>  Yunfei
>
>
>  ________________________________
>  zhangyunfei
>  2010-04-09_______________________________________________
>  clouds  mailing  list
>  clouds@ietf.org
>  https://www.ietf.org/mailman/listinfo/clouds
>
>
>  _______________________________________________
>  clouds  mailing  list
>  clouds@ietf.org
>  https://www.ietf.org/mailman/listinfo/clouds
>

--
Wang  Xu

_______________________________________________
clouds mailing list
clouds@ietf.org
https://www.ietf.org/mailman/listinfo/clouds