Fluent.Interface


Cloud computing report; Sun’s new API and DabbleDB

A recent McKinsey & Company report Clearing the air on cloud computing  reports that for companies to migrate their enterprise software platforms to the cloud:

The cost of cloud must come down significantly for outsourcing a complete data center to make economic sense

Although it is becoming cost effective for smaller ‘compute equivalent’ environment – where the total costs of assets for a typical data centre is US$45/month per CPU.

The report also suggests that:

CIOs should be working the cloud technology stack bottom up; cloud providers top down

Or in other words utilise virtualization and CDN’s available now, while waiting for the cloud service  providers to improved reliability and standardize offerings.  The recent Open Cloud initiative spearheaded by Sam Johnston from Australian Online Solutions might go someway to achieving this goal:

A common set of protocols and API’s for accessing cloud resources could be as revolutionary as HTTP was for TCP/IP.   And possibly the simplest and best implementation I have seen to date is Sun’s new cloud initiative, illustrated in this Hello Cloud example leveraging JSON over the RESTful API.  Hopefully the recent acquisition of Sun by Oracle won’t dent the enthusiasm of this project.

One of the great things about the cloud is the flexability developers/vendors have when choosing the software stack  for their implementation. DabbleDB was a recent example i came across that offers a rich database through a clever online spreadsheet-style interface.  It is built on the Squeak smalltalk engine utilising the SeaSide framework.  Avi Bryant talks to InfoQ on how his application works without the use of a RDBM instead storing virtual images of the smalltalk run-time.


CloudCamp in London

The good people at SkillsMatter recently hosted a series of lightening talks in their crypt in central London.

I can recommend checking out James Governor‘s blog which hosts a write-up by Benjamin Ellis.  Or watch the filmed video of the event.

Cloud computing, while the buzz at the moment is still in it’s infancy and this is particularlyobvious when looking at metrics around inefficiencies in power consumption required to run virtual machines.  But once this land grab stabilises expect to see standards emerge for accessing storage / data / virtual operating systems.


MPI.NET library enabling scientific cloud computing

The Message Passing Interface MPI standard to supporting writing software that run across many machines.  It has been used by the scientific community for high performance numerical libraries typically written in C or FORTRAN.

The University of Indiana recently released MPI.NET which requires the Microsoft Computer Server Cluster SDK on Windows 2003, or is naively supported on Windows Server 2008 HPC.

Cloud computing is shifting the way software is developed, so it probably won’t be long before native MPI implementations are supported in respective platforms such as Amazon’s EC2 with python, or naively on Microsoft’s forthcoming clustered offering.

Most software leveraging MPI such as PETSc is programmed in C.  Object oriented libraries have proved popular to programmers, and more recently functional languages such as Microsoft’s new F# (based on OCAML) have taken off as an alternative for mathematical libraries as seen in this blog post by Matthew Podwysocki.


Microsoft’s cloud computing offering with SDDS

Microsoft weighs in with its cloud computing offering in the form of SQL Server Data Services.  Currently in beta, users are able to sign up for free access.

A good screen cast walks you through the new tiered ‘ACE’ architecture which is designed around a top level ‘Authority’ a unit of geo-location which maps to an explicit DNS address eg:

brightsparc.data.beta.mssds.com

Under this Authority are Containers, and within a container is a series of Entities.  The flat structure is analogous to Tables and Rows, however a Row can represent different Kinds of data that doesn’t necessarily conform to the same schema. In fact any data other then the core ID, Version and Kind attributes are ‘Flexible Properties’ and set similar to a key based dictionary.

Under the hood this functionality could well be implemented with the new Filtered Index and Sparse Column features of SQL 2008 which allow for efficient storage of such data.

SSDS is a web service accessed over the wire using a SOAP or simple REST interface currently secured with basic credentials over https.  To get started I recommend downloading the SDK to create / update and query your data using the explorer GUI.  In REST speak, use POST to create a record, PUT to update and record and DELETE speaks for itself.  Moving forward cloud computing security is a challenge for Microsoft, but will no doubt tie in with it’s claim-based Card Space initiative.

Ryan Dun has a great blog with some examples of what is possible right now with SSDS, as well as very nifty open source REST library that can wrap POCO’s for persisting them to the cloud.

Although I did experience some unexpected outage on the service, Microsoft do plan to charge for this service in the future so I’m sure they will be looking at SLA of at least two 9’s if they are planning on competing with the likes of Amazon and Google.


Data Portability got a little eaiser with GNIP

A new data portability company GNIP has been getting a bit of press lately.  They are positioning themselves as the goto-infrastructure-publisher-guy for the social web.  A one stop shop where you can get pushed all your feed data from twitter, digg, etc… without having to ping every single provider.

They have timed their market entry well given Twitter’s very public performance woes.  But GNIP offers the possibility of returning aggregated data in a consistent way, so may help speed up the standardization of the various microformats out there on the social web.

I like GNIP because of it’s simple RESTful approach to design and architecture.  Their V1 was built in 90 days, and followed up with transitioning the system into the cloud care of AWS/Ec2 running on top of TerraCotta and ActiveMQ.  Very cool!