VOSA Architecture

VOSA is a web application.

This means that you only need a web browser (Firefox, Chrome, Explorer...) and an internet connection to use it. You don't need to install anything else.

All the VOSA operations are performed in the VOSA servers, no computing is done in your own machine. So you will only need the amount of memory that the browser needs to display the results (usually small) and the disk space to finally download the results files if you want to do it.

In exchange, as all operations are done in the VOSA server, big processes can be an important load to the server specially when several users are submitting jobs at the same time. And, in web applications, you usually have to wait, with the browser open, to the process to finish so that the results are loaded in the web page. If the process is long, this can be a problem.

Whats more, the advent of new and more sensitive surveys providing photometry at many wavelength ranges and covering large sky areas (GAIA, GALEX, SDSS, 2MASS, UKIDSS, AKARI, WISE, VISTA...) is pushing astronomy towards a change of paradigm where small groups, and not only large consortia, need to analyze large multi-wavelength data sets as part of their everyday work.

Thus, to be ready to work with large samples of objects, we have redesigned VOSA architecture since version 5.0 so that, keeping the advantages of being a web application, the drawbacks are reduced.

Now VOSA works with a distributed, parallelized and asynchronous architecture,with an improved design for large files. See more details below.

Distributed environment.

Now, most of the VOSA calculations are not done in the VOSA server. VOSA submits them to a different server and waits for the results. This reduces very much the VOSA server load, that does not depend much on the number of jobs or the size of user files.

In the future this infrastructure could be upgraded so that VOSA can distribute jobs among different servers to balance the load.

Parallelized computing.

VOSA is designed to work with files with lists of objects and make mostly the same operations to all of them.

But most VOSA calculations are now parallelized, so that it's not necessary to finish the work with one object to start the next one. The computation server organizes the jobs so that several of them are done in a parallel way, and collects the results once all the jobs are finished.

Asynchronous procedures.

VOSA communicates with the computation server in an asynchronous way. That is, VOSA submits a process and does not wait for it to finish. From time to time, or because a user requests it, VOSA checks the status of the process and, when it is finished, downloads the results, makes the final necessary processing and presents them to the user.

The main advantage of this is that you, the final user, do not need to wait, with the browser open, to the end of the process. You don't depend on the stability of your internet connection either. You can start a process, close the computer and come back later to see how it is going. If it is finished you will see the results. If not, you will see the status of the process and an estimation of the remaining time.

Processes can be canceled at any time from the VOSA web interface.

The asynchronous behavior is visualized in a similar way in all VOSA processes.

When you submit a process (for instance, a model fit), VOSA gathers the information, submits it to the computing server and waits for a while to check if the process is going to end shortly. So, during a short while (some seconds) you see something like this:

If VOSA sees that the process is not going to finish almost immediately, it tells you that the process has been submited asynchronously and gives you information of its status and (when possible) some estimation of the remaining time

At this point you can close the computer if you want and come later again to the page to see how the operation goes.

You can cancel the process whenever you want. You just need to click the "Cancel" button. VOSA will ask you for confirmation. If you confirm it, the process will be canceled and you can restart it again, maybe with different options. If you don't confirm, the process will continue as it was.

Once the process is finished, VOSA downloads the needed information from the computing server and starts processing it (what could include, for instance, making some plots).

And, finally, the results are displayed.

SVO Theoretical data server.

A great deal of the VOSA capabilities depend on what we call the SVO Theory Data Server.

• SVO Filter Profile Service: with properties of more than 3000 astronomical filters.
• Theoretical Spectra: more than 30 collections of theoretical spectra, with their corresponding SSAP services and some other web services for model analysis.
• Observational templates: 6 different collections of observational templates, with their corresponding SSAP services and some other web services for model analysis.
• Synthetic photometry for all the theoretical spectra and observational templates. This means more than 3 10⁸ values.
• Isochrones and evolutionary tracks: more that 50 collections and their corresponding web services useful for analysis.

VO services.

Another great part of the VOSA capabilities depend on external services, provided by different data centers and that are accessed by VOSA using Virtual Observatory (VO) protocols (ConeSearch, TAP and SSAP).

In particular, VOSA uses these services to search for objects properties (distance, extinction) and photometry.

SAMP.

SAMP is a VO protocol that allows to share data between VO applications. It was initially designed to work between desktop applications but, thanks to Mark Taylor's sampjs library VOSA, being a web application, can share results with desktop applications too.

Thus, apart from just visualizing and downloading the final results, most of the tables can be broadcasted to any other VO application that is open in the final user computer and connected to the SAMP Hub. In particular, this is specially useful to send some results tables to Topcat for further analysis.

See the SAMP section for more details.

Web design: pagination

For big user files containing thousands of objects, the plain visualization in a web page is not very useful. And, in some cases, it could even require a lot of memory and freeze or kill the web browser.

In order to avoid this problem, we have redesigned the presentation of large results tables implementing a customizable pagination form.

For instance, in the model fit, when there are many objects in the file you will see something like this:

In general, when there are many objects, you will not see the full list at once but only, for instance, the first 10 ones. And, together with that list, you will see a form where you can:

• Select an object so that VOSA directly goes to the page where the object is.
• Set the number of objects that you want to see in each page.
• Go to a certain page (or just the previous/next one)

Plots

Phys. Constants