From torrus

features for future development

WebUI Enchancements

The future WebUI will meet two general requirements:

1. More granular security
2. Decoupling the visual presentation from the datasource definition, and allow more dynamic content.

Date: Thu, 20 Jul 2006 15:49:19 +0200
From: "Christian Schnidrig" <[[mailto:christian.schnidrig@gmail.com|christian.schnidrig@gmail.com]]>
To: "Stanislav Sinyagin" <[[mailto:ssinyagin@yahoo.com|ssinyagin@yahoo.com]]>
Subject: Re: [torrus-devel] Torrus 2.x: the new WebUI
CC: [[mailto:torrus-devel@lists.sourceforge.net|torrus-devel@lists.sourceforge.net]]


Hi

My list. These are not all "must haves" of course, but I think a future GUI should at least not prevent the implementation of anything on this list.

Again I try not to think about how it could be implemented or how difficult it would be. It's not the time for that yet.
When I speak of bookmarks I do not mean browser bookmarks, but bookmarks managed by torrus itself. A view is not in the current torrus sense of view, but rather a page with one or more graphs on it.
The list is not in order of importance!

*) Changing view parameters should have immediate effect and not require lengthy and complicated update procedures before one does get to see what has changed.

*) Make gui independent of collector. With the current system I do not understand why I have to recompile and thereby interrupt the collection process simply to change the UpperLimit parameter of some graph.

*) Allow for a specialist (having deep understanding of the data collected) to create a fully customized view. One where he shows a list of handpicked graphs. This view should then be viewable by everybody provided they have the necessary access rights.

*) Maybe the views themselves should also have manageable access rights and optionally overwrite the graphs' access rights

*) It should be possible to add text into the views explaining to a viewer what he is seeing or give additional information .e.g. a view showing the number of customers checking in at the airport counter of company x might benefit from a text stating at what time there are scheduled departures for company x. Maybe that text should be imported from some other system in real time.

*) Custom views. A user should be able to browse through the data, put some graphs in a basket and then have the content of his basket displayed on one single page. That page should be bookmarkable and the user should be able to make his view accessible to other users as well.

*) Using the same basket concept. The user should be able to display the sum, max, min, etc of all graphs in his basket.

*) If a user finds some interesting data. he should be able to bookmark the graph, the time, and the temporal resolution at which the specific anomaly shows best.

*) Users should be able to annotate graphs.

*) Have a time and date picker. The GUI should not need to reload the page simply to change the displayed day, thus allowing to quickly cycle through several days of data at high temporal resolution.

*) It should be possible to drill down any graph in any view.

*) when showing a view, let the user choose the time window (day, month, year, etc). Network traffic may be best viewed in a daily graph but medical data like a patients temperature is probably better viewed over a longer time range. Therefore torrus should not force a particular time window on the user. And it's not good enough to let the user drill down. One cannot drill down every graph and view them next to each other at the same time.

*) Upper & Lower limit should be context sensitive or defined by the view. It is very hard to compare graphs if the scale is not the same for all of them.

*) It would be cool if horizontal lines could be defined per view and then be supper imposed on all graphs. The ability to superimpose a graph would even be hilarious.

*) Hobbit allows for a user to select a time interval on a graph. The view is then zoomed in to that very time interval.

*) Allow for displaying MIN,AVERAGE,MAX values. Each one by itself or together in a single multigraph

*) Dynamically adapt the views to the window size of the browser.

*) let the user choose the size of graphs

*) The ability to define a default view for a given type of entity. e.g. when you hit a host then show me CPU load, CPU usage, Memory usage, and disk space in one page.

*) when showing a cdef, provide a link for a view where all graphs the cdef is composed of are displayed next to each other. This helps a lot if one wants to find out where an anomaly originates from.

*) if a graph is also part of some other view. Provide the links to all those views whenever the graph is displayed somewhere.

*) when displaying a graph give a list of links to graphs which are related. What related means should be configurable and independent of the data of course. The relationship may come from an external system in real time.

*) Allow for direct links to the graphs bypassing the gui. That's needed to integrate torrus seamlessly into some other system.

looking at google mail:
*) gmail does not store emails into directories. All you can do is give emails so called labels. Every email can have several labels.
*) if you know something of the email your looking for, simply search for it. You don't need to know what label it has, respectively in which directory you'd find it.

Regards
Christian