High performance chart for a lot of data in javascript

Question

I need a high-performance charting component. The main technology used is not specific; maybe javascript, flash or java for what it costs.

I need to solve these problems:

1. I need to determine the correct chart type for this type of data:

001. testA 002,003 002. testB 003,004 003. testC 004 004. testD 

I am currently thinking of a network view:

http://www.connectedaction.net/wp-content/uploads/2009/09/2009-September-NodeXL-CHI-2010-Tag-Network.png

http://visualgadgets.blogspot.com/2008/06/graphs-and-networks.html

2. The schedule should exceed performance:

To date, I have over 604 million records, and I expect it to grow to a billion in the next few years. This is a lot for visualization.

3. The chart must be accessible from javascript

Given problem 2 , I will need to visualize the chart at a time (during the sartup application) and then update it gradually. By "update" I mean "add records"; records (and links between nodes) will not be deleted.

---

I don’t know what else I could add, that’s almost all.

---

Edit: I got some useful advice; I could use pre-processed data to show a complete graph, and then only show real-time data when each of them scales accordingly.

However, given my data format above, this is a bit difficult to do, given that there may be circular links. I have to think about it.

In the meantime, I still need advice on a good diagram component for network / topology diagrams.

Answer 1

This eliminates the mapping of servers, since your server will have a lot of resources if you maintain a site with high traffic. It would be logical to allow the client side to process the chart.

If you do not show many diagrams on one page, the performance should not be as large as the client side of the transaction. However, to use the data, you will need to reformat it for ANY toolkit for creating diagrams / reading frameworks.

Raphael is a good visualization toolkit, there may be tools that expand Raphael to make a more specific schedule. Afaik, you need to do a lot of manual coding yourself when using Raphael out of the box (although still only once for all of your billions of diagrams, oP).

Here is a link to a list of top 5 jQuery diagrams.

More can be found with a short google session;)

Answer 2

Another opportunity to think about is a preprocessed database. I did this a few years ago on the sql server, we created a second database server with some very complex DTS packages and sql scripts that took up a huge amount of data that we worked with and summarized them so that the Crystal report server processed the reports.

you lose the ability to receive reports in real time, but if this is not a requirement, it worked well for us, and you reduce the load on any report server that you use and reduce the load on your production database to run reports and let it concentrate on starting applications, it also reduces the load on your client browsers, because you do not need to send your chart to this huge piece of data.

as far as visualizing, I've heard a lot of good things about JavaFX and its charting capabilities.

Answer 3

Regarding the charts, I found the following candidates:

ProtoVis and jsViz

Both of them are based on javascript, so it’s pretty slow when rendering (I think?). I think I will try ProtoVis first.