[tsc] A little help needed

Maguire_Tom at emc.com Maguire_Tom at emc.com
Wed Nov 22 05:59:31 CST 2006


My $0.02USD

Tom

_______________________________________________
>Folks on the TSC,
>
>Mark needs a little help, and on a short fuse.
>
>I have tried to outline some simple ideas to these questions, but I  
>think David B. needs more detail than I can provide at present.
>
>On 21 Nov 2006, at 18:38, Linesch, Mark wrote:
>
>> David, Joel,
>>
>> Need a little help from the Standards/Technical team here.
>> In the process of writing the Grid and Distributed Computing  
>> Landscape document, David Birnbaum has come up with a series of  
>> questions that need to be answered for the reader. Can you look  
>> these over and do the following: (a) Answer these questions from  
>> your perspective; (b) identify an OGF expert who might be able to  
>> answer one or more of these questions and send them the question  
>> and solicit them to provide a response. The responses can be in the  
>> form of a short paragraph per question but try to be as specific  
>> and factual as possible in your answer. David will utilize your  
>> responses and others as in-depth source material as he continues to  
>> write the document.
>>
>> QUESTIONS:
>> 1. What are the issues or limitations of a virtualized server  
>> infrastructure that can be addressed by grid capabilities?
>Virtualizing a server infrastructure only addresses one layer in the  
>system. The Grid provides, using many of the same concepts as SOA,  
>virtualization of higher layers of the stack, e.g. application etc.

>> 2. What are the issues or limitations that might be encountered  
>> during either a pilot or production implementation of SOA that can  
>> be addressed by grid capabilities?
>The key issues here are scale and dynamics. Most SOA deployments are  
>relative small scale of highly specialized. Adding Grid to SOA  
>provides a more dynamic platform, allowing new service to come and go  
>continuously.
>> 3. Applications
>> a. Which non-compute intensive applications are well suited to  
>> running on a grid?
>This one is out of my scope.

Business intelligence/analytics.  These applications while actually
'compute' intensive are not what most consider w.r.t. that term.
ETL processes are also well suited to grids.

>> b. What modifications are needed to run an application on a grid?  
>> Does the application have to support parallel processing?
>If the aim of using Grid is to exploit many servers on a single task,  
>then yes parallelism is necessary. This does not necessarily mean the  
>application needs adaptation. Many applications are already parallel  
>in nature or the parallelism can be supported by parts of the Grid  
>middle ware. For example, most transaction based applications  would  
>not require changing, provided that the underlying database provided  
>support for distributed transactions.

Absolutely agree, in fact most 3-tier applications are incredibly
well suited to run on a grid.  The noticeable exception is the final
tier (database).  Only some vendors have database assets that work
well on a grid.  Even those have difficulty with the dynamic 
provisioning aspects of grid infrastructure.

>> c. Are I/O intensive applications not suitable for grid?
>Provided that the infrastructure and application are designed for it,  
>data intensive applications work very well on a Grid. care is needed  
>in ensuring that data transfer bottle necks are managed and that the  
>data itself is distributed across the Grid infrastructure. The  
>applications should be configured to tolerate latency, e.g. using  
>work queues.

Not sure I agree with the last statement.  There are many new techniques
for building highly scaleable data grid infrastructures absent latency.
The real question is access to the data parallel or serial and that is
an application design issue.

>> 4. For an enterprise IT architect, what grid standards, services  
>> and technologies does OGF recommend that they have on their  
>> planning horizon over the next three years?
>The list in the Technical Strategy Document is a good start. I would  
>add to that standards for data centre management, which may come from  
>OGF or elsewhere.

I would add data center fabric standards, semantic web standards,
storage virtualization standards off the top of my head.

>> Please email your perspective/response to these questions directly  
>> to David and copy me by EOD, Wed, November 29th. Your help is much  
>> appreciated.
>I know this is less text than you need, but at least it's on time. I  
>hope the rest of the TSC can help.


-- 

Take care:

     Dr. David Snelling < David . Snelling . UK . Fujitsu . com >
     Fujitsu Laboratories of Europe
     Hayes Park Central
     Hayes End Road
     Hayes, Middlesex  UB4 8FE

     +44-208-606-4649 (Office)
     +44-208-606-4539 (Fax)
     +44-7768-807526  (Mobile)






_______________________________________________
tsc mailing list
tsc at ogf.org
http://www.ogf.org/mailman/listinfo/tsc

-------------- next part --------------
A non-text attachment was scrubbed...
Name: smime.p7s
Type: application/x-pkcs7-signature
Size: 3654 bytes
Desc: not available
Url : http://www.ogf.org/pipermail/tsc/attachments/20061122/75a70364/attachment.bin 


More information about the tsc mailing list