meteohub.de

Let's make use of the feature of Version 3.2d to report recomputation speed for you Meteohub system. Beside the platform itself, I will expect the storage media to have a major impact as well. So please tell us your...

• processed data records per second (from "meteohub log" )
• platform (NSLU2 133MHz, ALIX.1D, etc...)
• storage media (type of USB Stick, SSD, CF card, etc)

• 14612451 records processed in 7265 seconds (2011 records per second)
• fit-PC Slim
• Transcend SSD TS8GSSD25-S

This is really fast compared to my NSLU2 experience ;-)

Meteohub on my server's VMware is even quite a bit faster :-)

• 14153300 records processed in 766 seconds (18476 records per second)
• VMware Server 2.0, Core2Duo, 3GHz

Here good old NSLU2 as reference

• 14179626 records processed in 29704 seconds (477 records per second)
• NSLU2 266 MHz
• 4GB USB stick OCZ ATV

• 7877937 records processed in 21277 seconds (370 records per second)
• NSLU2 266 MHz
• 4GB USB Freecom Databar

Data nature: 2x anemometer (14 second update), 2x rain (~45 second updates), 7 temp/hum or temp only (~ 1 minute updates). Most sensor in place for ~9 months.

Load during recompute: Reading from USB RFXCOM, updating one html template @ 15 minute intervals, feeding WD by TCP/IP protocol, another Meteohub (ALIX 1D) observing port 5557.


Memory age: 9 months

• 7879599 records processed in 4193 seconds (1879 records per second)
• ALIX 1D
• 4GB iCF 4000 Dual Channel Industrial Compact Flash - 266X

Data nature: 2x anemometer (14 second update), 2x rain (~45 second updates), 7 temp/hum or temp only (~ 1 minute updates). Most sensor in place for ~9 months.

Load during recompute: Reading from NSLU2 Meteohub by TCP/IP, updating one html template @ 15 minute intervals.

Memory age: New

# 5371448 records processed in 5021 seconds (1069 records per second)
# NSLU2 266 MHz
# Corsair USB Flash Voyager 4GB

Looks like the memory makes a difference. :)

pirelli wrote:Looks like the memory makes a difference. :)

Seems likely.
I just edited my posts to add in a few more details (nature of data, load during recompute, and age of the memory device) in case that helps build a better picture.

It might be interesting for me to try again after I have switched the ALIX 1D over to being the main device (i.e. after swaping the "load during recompute" characteristics from NSLU2 => ALIX 1D and vice versa).

- 912268 records processed in 741 seconds (1231 records per second)
- eBox 4300 (eTC-2500), 512MB RAM
- 4GB CF

skyewright wrote:It might be interesting for me to try again after I have switched the ALIX 1D over to being the main device (i.e. after swaping the "load during recompute" characteristics from NSLU2 => ALIX 1D and vice versa).

The figures for the ALIX 1D as "main device" (i.e. with a higher load during recompute) were almost identical at 1804 records per second.
I can't compare against the NSLU2 as "secondary" at present, because the file system on its USB stick failed and I haven't yet rebuilt it...

Version 3.3 report.

recomputation of weather data finished: 145206 records processed in 84 seconds (1728 records per second)

system'ALIX 1 d

4 G. card

Update time every 10 minutes by cresta wxr 815

Power consumption 5,3 watt maximum !!

Regards
Paul

[*]4972334 records processed in 2604 seconds (1909 records per second)
[*]Alix.3C2
[*]4GB NoName CF-Card

I see you have it running on ALIX.3C2. Very interesting, as I have not tested this. Difference between C and D models does not seem to be to much, but anyway good to know.

I ordered Alix3D2 but was supplied with Alix.3C2 (Varia-Store). The 3C2 seems to be the predecessor of 3D2 but works fine so far. The MeteoHub System Info indicates a 3D2, but a print on the circuit board and the invoice from Varia-Store tell me its a 3C2.

- V3.3
- 3152527 records processed in 1377 seconds
- (2289 records per second)
- fit-PC slim with 4GB Transcend SSD

Regards
Juergen