a journal of a researcher

Thursday, November 08, 2007

Globus Tools Updates

Some facts about Globus toolkits. You can see I am working on some real world projects now.

The role of the Globus Toolkit, as in its own document, is to produce general solutions to problems (design patterns) that could be reused from application to application. The components in the toolkit are standard building blocks and tools for application developers and system integrators. The Globus toolkit does not provide a “turnkey” solution for any project or product.

The importance of WSRF in Globus Toolkit

Globus sees a Grid Web service is a WSRF-compliant WS which has properties defined in a property xml, and its WSDL has a port to download the property. You can subscribe a property and get notification when a property is changed. The WSRF framework effects how a WSRF-compliant WS is indexed, discovered, monitored and invoked.

The toolkit is a collection of solutions to sub-problems in grid computing. List of Globus Toolkit components:

  • Two SDK
    • Web Service Core implementation
    • Grid Security Infrastructure (GSI) implementation
  • A variety of basic grid services
    • Computing/process Scheduler (GRAM)
    • Data Management (GridFTP, DAI, RLS)
    • Monitor/Discovery (MDS)
    • Authorization/Security(CAS)
    • In development Telecontrol (NTCP/GTCP), Metadata(MCS), Virtual Data (Chiera, Pegasus)
  • Developer APIs
  • Tools and Examples

Compliant to the following standards

  • SSL/TLS v1 (from OpenSSL) (IETF)
  • LDAP v3 (from OpenLDAP) (IETF)
  • X.509 Proxy Certificates (IETF)
  • SOAP (W3C)
  • HTTP (W3C)
  • GridFTP v1.0 (GGF)
  • OGSI v1.0 (GGF)
  • DAI (GGF)
  • WS-Agreement
  • WSDL 2.0 (W3C)
  • WSDM

Tuesday, November 06, 2007

A Mini Report on Amazon EC2 and S3 Web Services


You can run a virtual image of your OS in Amazon Elastic Compute Cloud (EC2)

You can storage any amount of data using Amazon Simple Storage Service (S3)


EC2 and S3 have Web service interface (SOAP and REST) and they are called Web services by Amazon. Be careful, EC2 assigns dynamic IP. You can also use command line to operate EC2, but not for S3 (seems). S3 is designed for files. You can put a database on EC2, not on S3.


S3: Storage ($0.15 per GB-month), Data Transfer ($0.10 per GB transferred in and $0.18 per GB transferred out … ), Requests ($0.10 per 1,00 PUT or LIST requests …).

EC2: $0.10 for a small instance-hour, Data Transfer ($0.10 per GB transferred in and $0.18 per GB transferred out … ),

Prices Comparison:

If you host a Web site as a small instance in EC2, the basic cost is $72/month. If you have a large amount of traffic for your Web site, you need to throw in more money. For example, 1TB is about $200.

Yahoo can host a small business Web site for $20/month. So, if just hosting a Web site, EC2 is not cheap. But $72/month is relatively the market price of renting a Web server.

Here is an article (http://www.webpronews.com/node/32013/print) for replacing a home backup server with S3. If you count in how much you spend on a backup server, software, electricity in 5 years, you save money and maintenance efforts by switching to S3. Be careful, this article is about backup data, so the traffic cost in this case is minimal.

Some analysis on benefits and when to use:

It seems that the price for storage is quite cheap, but the price for traffic is not cheap. As EC2 assigns dynamic IP, it seems it is not good for hosting a Web site. But some people do agree that the application is more stable than you expected, i.e. you can keep it running forever.

The best thing for EC2 and S3 is the scalability. You can easily increase the capacity of your service by starting more instances at Amazon. There is no limitation on how much resources you can use. And the set up time is very short. Other benefits are reliability, security etc. You can save money by avoiding setting up your own information infrastructure, if you have serious applications. Another way is to use Amazon to deal with peaks, as you can easily switch on and off the instances and you pay according to how much you used.

Monday, November 05, 2007

2008 IEEE Services Computing Contest is announced today

I am chairing 2008 IEEE Services Computing Contest. After several months of consulting and preparation, the Contest is officially launched.

The objective of Services Computing Contest is to promote research and teaching services computing techniques in universities and colleges worldwide. This contest brings together talented students to solve challenges in using Service Computing techniques to build value-added business systems in the era of Web 2.0. Following the two successful contests in 2006 and 2007, the 2008 IEEE International Services Computing Contest has the theme of "Business SOA and Services Mash-up" that the participating teams will utilize the services and data from different sources and integrate them into a new value-added business service to the end users. Each team will submit a paper and build a live system. The finalists will be invited to either the SCC 2008 (Hawaii, USA) or the ICWS 2008 (Beijing, China), at their will, for the final competition. The reasons to organize the final competition in two conferences are to encourage teams from different countries to participate and to solve practical traveling issues. The contest is one competition. The finalists will be announced in the SCC 2008 and the winners will be announced in the ICWS 2008. The papers of the finalists will be published in the proceedings of the 2008 IEEE Congress on Services (SERVICES 2008). This contest is sponsored by IEEE Computer Society Technical Committee on Services Computing, and is supported by IBM Research (Services Computing).