[Myexperiment-discuss] Re: Web service metrics: interface granularity and service composition

[Michael Parkin]

Thanks Duncan.

Had a quick scan of the paper you linked to. It seems the traffic  
light system is concerned with the health of the deployed service.  
Apologies, perhaps I wasn't too clear about what I'm looking for, by  
granularity I meant two things:

* Granularity of a service interface (or the 'size' of a service  
interface). Aka how many operations/portTypes/interfaces/etc. in each  
interface.

* Granularity of a service operation (the 'size' of a particular  
operation). A service operation can perform a single function or more  
than one function. Singular operations can be, for instance,  
primitives like CRUD while composite operations combine other service  
operations.

Our research is concerned with finding the appropriate granularity for  
a service interface and each operations when designing or refactoring  
a service, according to some criteria like reusability, value,  
maintainability, reliability, etc.

Initially we're attempting to find some data - such as number of  
operations per service interface - from existing service interfaces to  
find some real-world data on the first aspect of granularity. Hence  
our interest in the Biocatalogue, myExperiment and others, such as  
EMBRACE and seekda.

Thanks again,

Michael.

p.s. I used 'composition' a bit too loosely in the last email - I  
meant the composition of the service interface wrt operations, not  
compositions of services :).

On 16 Jul 2009, at 14:23, Duncan Hull wrote:

> Hi Michael
>
> Good to hear from you.
>
> On 16 Jul 2009, at 12:33, Michael Parkin wrote:
>
> > Hope you are well and (still) enjoying life in Manchester. Looking  
> > forward to the new season without Ronaldo but with Michael Owen? ;)
>
> Good riddance ronaldo! And Tevez has switched from red to blue... so  
> we won't be saying goodbye just yet.
>
> > Enough football though: I've a question about the EBI's  
> > Biocatalogue and don't know where to start. I'm hoping you can help  
> > as you're familiar with the community.
> >
> > As part of my work here I'm looking into metrics of services, like  
> > the interface granularity and composition, since there is little  
> > published data about this. I'd like to ask the EBI/UoM if they have  
> > this (or any other useful) data on the service descriptions they  
> > host, or if they could generate it from their repository. (ok, I  
> > could go through all the interfaces myself and get the data, but  
> > really don't fancy it.)
>
>
> Not sure what you mean by granularity.... but there are metrics on  
> use (green, amber and red) some of which are described in this  
> paper: http://dx.doi.org/10.1093/bioinformatics/btp329
>
> > I see that most entries in the bio-catalogue just contain links to  
> > the WSDL, API docs, etc. so it may be that they don't have this  
> > information (it would be interesting to find out what the EBI  
> > stores about these services).
> >
> > So, my question is really simple: who should I contact?
>
> I've cc'ed this message into myExperiment and biocatalogue mailing  
> lists who should be able to provide more info.
>
>
> > Supplementary: does myExperiment collect similar data? I understand  
> > they have a more involved curation process and look at the services  
> > in more detail.
>
>
> Yes, AFAIK myExperiment keeps a record of which services have been  
> used together - someone from the myexperiment team could probably  
> supply more details if you are interested.
>
> Hope this helps.
>
> Duncan
>
> ---
> http://duncan.hull.name