Web upload size

[Stuart2007]

Can anyone give me a guide to the typical basic cumulus webpage upload sizes, assuming the standard templates are used? I assume some pages are not uploaded every time (records, yesterday etc). Not sure how much is uploaded for gauges...

I am going to have to use 3G for the internet connection and wonder how much space will be needed each month (probably say 5 minutely updates).

I've searched here in vain for 'upload size', so I hope I have not overlooked the information in an obvious place.

Thanks

StuartM

[steve]

All of the pages and images are uploaded on each update. In total, it's about half a megabyte - about 100KB for the pages and 400KB for the images.

[water01]

I do not think there is an exact measurement as everyones T (template) files will differ according to their modifications.

You are incorrect in assuming that records, yesterday etc. do not get uploaded, all T files get uploaded on each upload.

I have done a rough calculation based on my T files and the Trend .png files that are uploaded each time and it would appear to be about 523K is uploaded as standard (including realtime.txt). But as I said that is based on my files so it could differ.

[Stuart2007]

All of the pages and images are uploaded on each update. In total, it's about half a megabyte - about 100KB for the pages and 400KB for the images.

Thanks Steve. Rather than reducing frequency to get below a 3G upload limit, I suppose we could forgo some of the upload files (eg graphs and gauges) and look for another way to give a simple analog web view of the data, possibly using javascript for example...?

StuartM

[steve]

Yes, the major saving would come through omitting the images. The gauges page itself is quite large, even ignoring the images, because it contains a lot of data for the wind distribution gauges. Omitting that as well gets you down to about 45k per upload (in 1.9.2, at least. Version 1.9.3 has the monthly records page which contains a lot of data, and is 25k).

You could replace the standard gauges with the SteelSeries gauges, which just require a small data file to be uploaded, and are far superior to the standard gauges anyway.

You could also reduce that 45k figure still further by moving to PHP pages, so the pages themselves don't get uploaded, just the data that they use.

[Stuart2007]

Yes, the major saving would come through omitting the images. The gauges page itself is quite large, even ignoring the images, because it contains a lot of data for the wind distribution gauges. Omitting that as well gets you down to about 45k per upload (in 1.9.2, at least. Version 1.9.3 has the monthly records page which contains a lot of data, and is 25k).

You could replace the standard gauges with the SteelSeries gauges, which just require a small data file to be uploaded, and are far superior to the standard gauges anyway.

You could also reduce that 45k figure still further by moving to PHP pages, so the pages themselves don't get uploaded, just the data that they use.

Thanks Steve, I was indeed thinking about the Steve series gauges. 45k etc would be fine.

StuartM

[mcrossley]

Attached is a modification of the standard monthlyrecordsT.htm that reduces the processed file size slightly, down to 21KB (every little helps! 4KB per upload in this case) by reworking the JavaScript.

monthlyrecordT.zip

[sfws]

Following on from what Mark has suggested, some sites save a little bit more by combining pages, they only need one set of headings and labels for the two sets of values.

(This assumes ability to use a suitable editor to merge elements from the two pages, and some understanding of HTML to recognise start and end of elements and rules like each row of a table having same number of cells).

There are sites with yesterday/today combined, monthlyrecords/records combined, and/or thismonth/thisyear combined, on pages otherwise very similar to original Cumulus ones.

I'm not at my own PC, so cannot look up my bookmarked examples, but maybe someone who has combined pages will respond with an example containing an example unprocessed Cumulus template.

[mcrossley]

Err, mine? Well today/yesterday, and month/year at any rate.

[sfws]

Err, mine?

Sorry Mark - I was only thinking of you re JavaScript and Steel Series, now I will make myself a note about what you have on your site ready for any future time. Anyway one site that inspired me (and I could not remember in my last post) was http://www.cheadlehulmeweather.co.uk/today.htm.

Anyway Mark's website http://weather.wilmslowastro.com/today.htm shows that each table data row is in this format (if today preceeds yesterday):

Code: Select all

<tr class="td_temperature_data">
      <td>High&nbsp;Temperature</td>
      <td><#tempTH> <#tempunit></td>
      <td>at <#TtempTH> </td>
      <td><#tempYH> <#tempunit></td>
     <td>at <#TtempYH></td>
</tr>

i.e. each data row is constructed from the row in today with the last two cells from the yesterday row appended. The heading rows can be seen by viewing source in the online HTML.

[Ned]

A bit off topic, but I've long wondered why the upload takes so long (1m 50s in my case) Is this typical? Or could it be an FTP server bottleneck?
Speed testing reports my upload speed is 0.8Mbps while emailing a 500kb file takes about 5 secs.
Cheers.

[steve]

There's a protocol overhead with logging in, changing directory etc, and with each file uploaded, and the standard web site contains about 40 files. If you turn on ftp logging you'll see this. For me, the upload takes nearly two minutes with a 448 kbps upstream connection speed, but the actual transfer speed isn't a huge factor; latency has a bigger effect.