All new Registrations are manually reviewed and approved, so a short delay after registration may occur before your account becomes active.
ReliableSite.Net's Quad Core ARM Server $19/m
Don't recall which was the original thread where the offer was brought up but here's a benchmark report:
http://serverbear.com/benchmark/2013/05/03/FriPHG6KiZTCTk4h
I did get it knowing that it was going to be a "for fun" server because I was curious of its performance. Now, I kinda wish I could get my $19 back
The hardware is indeed a ODROIDU2. I also ordered it with a 64GB SSD. It wasn't mounted when I ran the test, so the I/O performance you see there is probably from reading an SD card.
After I partitioned and mounted the SSD:
root@xxx:/home# dd if=/dev/zero of=io-test bs=1M count=1k conv=fdatasync
1024+0 records in
1024+0 records out
1073741824 bytes (1.1 GB) copied, 45.6014 s, 23.5 MB/s</code>
Main -> SSD
root@xxx:~# dd if=/dev/sda1 of=sb-io-test bs=1M count=1k conv=fdatasync
1024+0 records in
1024+0 records out
1073741824 bytes (1.1 GB) copied, 99.2937 s, 10.8 MB/s
SSD -> Main
Not that amazing anyway. I can't be bothered with the other read/write tests.
I thought the server would be at NYC (what the WHT post said and also why I jumped on the offer), but the traceroute says it is out of Choopa in NJ.
Sales was immediate in my initial emails until money exchanged hands last Friday. Was aware of the 3-5 business days delivery time. Emailed twice (once on Tues and again on Thurs) to inquire on status - do note that it was a polite email asking for an updated ETA because I'd like to know when I can expect it to be delivered (i.e. it wasn't a "Where is my server?!@#" email). No response received to either emails.
Finally, on Thurs (yesterday), I decided to open a ticket and said that if they don't respond to these simple inquiries, I'd like my money back. Response was a minute later saying that it'll be up by Friday at the latest. Server was delivered around midnight the same day.
Comments
They're located in NYC, but their datacenter is in NJ I believe.
From: http://www.webhostingtalk.com/showthread.php?t=1262341
Pardon my ignorance, but looking at the map Matawan, NJ looks VERY close to NYC.
w..why?
Good to know I doubt they would shell out an extra $46 for the odroid X2..
you should do the DD test on the USB drive the results for the DD test done on the main flash drive would definitely give you a very low DD result cause sd cards have a very bad write speed..
Choopa is 38 miles from the center of NYC or 14 miles from NY. Literally in the grand scale of things there's so little distance that it doesn't matter.
And I love the choopa network. I have about 20 servers in their network and it is one of the best ive used
Interesting to note that the UnixBench score is about the same as on Atom N2800.
I/O is of course slower, maybe SSD is connected via USB? You can check in "lsusb", or even better, pastebin us the whole "dmesg".
That benchmark says: CPU Speed 798 MHz. Maybe a downclocked Atom?
Pretty sure those are ARM CPU's.
@Alex_LiquidHost i meant the link to the Atom benchmark that @rm_ has in his comment.
@BlueVM: True that.
I'm Canadian so I'll put it this way. It's like telling ppl your server will be in downtown Toronto but in reality, it's in Brampton. Then say it's still GTA (Greater Toronto Area).
Trust me, it's different.
Anyway. I'm not going to fight over the nitty-gritty about network. It is, after all, just $19. I just wish they were upfront on the location.
Probably that one had CPU frequency management enabled (downclocks when idle, and that happened before start of the benchmark when the CPU frequency was recorded), I changed the URL in my post to point to a normal-frequency benchmark result, which is still about the same.
Who really cares though? Network is about connectivity, not physical location. If the connectivity to NYC at Choopa is not substantially similar to what you would get right in NYC, then there would be something to complain about.
@rm_: Have to focus at work so here, have fun: http://paste.ubuntu.com/5629687/
Indeed, the SSD is attached via USB, that explains the low dd result.
Does anyone have the benchmark for the cubieboard? It has slower CPU and less RAM, but also has a SATA port which should speed up the IO, at least on theory.
@rds100 Mele A2000 (same hardware as the Cubieboard, but with less RAM and a 25% lower RAM frequency) with a Corsair F60 SSD plugged into SATA:
I did not run a UnixBench on these, but from googling found someone reporting a result of 230.
ODROIDU2 has only usb ports, no sata..so the disk can only be connected through usb.
mobilegeeks.de/wp-content/uploads/2012/11/odroid-u2-1.jpg
@rm_ thanks! Not that bad, could make a nice lowend storage server from one of these.
So, getting it with the ssd is useless.
@rds100 People make web server clusters out of them http://guillaumeplayground.net/allwinner-a10-cluster-mele-a2000/
though I guess that one is more for fun and experimentation, all things considered it would not compare too well with a "traditional" 8 or 16-core server.
I imagine random I/O would be better than the spinning disk.
Good point!