小撒，翻译一下，谢啦！

谁帮忙翻译下这篇文章，谢啦！

Enterprise Grade Cloud Computing
Paul Murray
HP Labs
pmurray@hp.com
ABSTRACT

Cloud computing and the as-a-service paradigm have gained a lot of  interest  recently.  The  separation  of  service  provider  from infrastructure provider has made  it much  easier  for new  services to be established online quickly and with low financial risk, and to scale  those  services as demand dictates. They can be built  to  run
directly  on  top  of  infrastructure  such  as  Amazon  EC2  [1],  on application  platforms  such  as Google App Engine  [3], or within higher  level  platforms  such as  FaceBook  [2]  or  force.com  [6], with  increasing  levels  of  ease  of  development  and  task specialization.

It  is clear why a startup company might be attracted  to  the cloud computing model. Equipment  is very  costly  to purchase  and  can only  be  amortized  over  a  period  of  years. Using  someone  else’s infrastructure on a pay-per-use basis converts this fixed cost into a variable  cost  based  on  actual consumption;  reducing  initial investment  and  risk. Also  the demand  for online services  can be very  variable  and  poor  response  due  to  overload  can  risk losing customers. So the ability to scale the use of cloud compute power also mitigates the risk of failure.
The  arguments  for  an  established  enterprise  are  not  the same. Such a business would have a well understood compute capacity and  multi-year  investment  lifecycles.  As  the  financial  risk becomes  less  significant  other  issues  come  into  play  such  as security,  legislation,  and  dependence  on  the  provider.  Exactly where  data  resides  is  important  as  it will  be  accountable  to  the local  legal system,  especially where  the main  line of business  is concerned.  Security requirements  may  not  be  compatible  with those offered by existing infrastructure providers.

In reality enterprises already use a mix of services, some in-house, some contracted out. They may use a pay-per-use model to access outsourced  payroll,  travel arrangement,  or  even  legal  services. These  may  be  provided  by  a  cloud computing  platform  and integrated with  further  services  such  as  credit  card payment  and courier distributors. 

For  in-house services  there may still be a cost advantage in using someone  else’s  infrastructure  –  if  they  are  big  enough.  Studies have  shown  that  the proportional  cost  of  building  and  running data  centers with  tens  of  thousands  of machines  is  significantly lower than one with just a few hundred. Where an enterprise does retain  its own internal systems for IT or its main line of business, it  is  likely  to  be  interested  in  using  a  private  cloud;  its  own
 
Permission to make digital or hard copies of all or part of this work for personal or classroom use is granted without fee provided that copies are not  made  or distributed  for  profit  or  commercial  advantage  and  that copies bear  this notice  and  the  full  citation on  the  first page. To  copy otherwise,  or  republish,  to  post  on  servers  or  to  redistribute  to  lists, requires prior specific permission and/or a fee. WDDDM’09, March 31, 2009, Nuremberg, Germany. Copyright 2009 ACM 978-1-60558-462-1/09/03…$5.00. internal  infrastructure and services managed  in  just  the same way
as  an  external  cloud  provider  would  do.  But  they may  also  be interested expanding out  to external clouds to accommodate peak demands. An enterprise is likely to straddle the line between self-owned,  on-premises  compute  facilities  and  third  party  cloud infrastructure. 

The  future will bring a world of services  interacting securely and running across multiple infrastructures, scaling and distributing as required.  These  considerations  provide  drivers  for  cloud computing research, both at service level and infrastructure level. 

At  HP  Labs  we  have  been  investigating  service  provision  in  a shared  compute  infrastructure  for  more  than  a  decade.  Past prototypes  include  SoftUDC  [5],  Frame  Factory/SE3D,  and  the HP Utility Rendering Service  [4] used by DreamWorks  to create the  Shrek  and  Madagascar  movies  among  others.  During  the course  of  this work we  have  addressed  a  variety  of  challenges. How  can  clients  with  low  bandwidth  connectivity  interact effectively  with  a  service  that  involves  very  large  volumes  of data? How can independent services operate, flex and scale within
the  same  shared  infrastructure  yet  achieve  sufficient  isolation  at the  data  and  service  levels?  How  can  data  and  services  be managed across multiple geographically distributed data centers?

In addition  to  the  technical challenges we also experienced client behaviors during  these  trials  that have both motivated  the use of cloud  computing  for  individual  users,  but  at  the  same  time challenged the economics of the whole paradigm. 

Our current  research addresses  the challenges of enterprise-grade cloud  computing,  starting  with  the  question:  what  would  cloud computing need  to provide  for  enterprises  and  enterprise  service providers?  In  this  talk we  discuss  our  view  of  cloud  computing and  what  we  are  doing  to  address  the  challenges  of  this  new paradigm.
1.  REFERENCES
[1]  Amazon. Elastic Compute Cloud. http://aws.amazon.com/ec2
[2]  FaceBook. http://www.facebook.com/
[3]  Google. App Engine. http://appengine.google.com/
[4]  Hewlett-Packard. Servicing the Animation Industry: HP’s
Utility Rendering Service Provides On-Demand Computing
Resources, 2004. http://www.hpl.hp.com/SE3D
[5]  Kallahalla, M., Uysal, M., Swaminathan, R., Lowell, D.,
Wray, M., Christian, T., Edwards, N., Dalton, C., Gittler, F.
SoftUDC: A Software-Based Data Center for Utility
Computing. IEEE Computer, 37(11):38–46, 2004
[6]  Salesforce.com. force.com. http://www.salesforce.com/force