Avoneer
|
| posted on 26/3/06 at 10:43 PM |
|
|
Wireless ???
I'm moving into a new house in a couple of months and will be getting a laptop as well as my desktop computer.
I know the easiest way to wire them all up will be to go wireless.
My only concern would be the download speed compared to a hard wired set up.
How much slower would it be?
On a good day, I can download at about 60 B/s normally so what would this be like on the computers once they go wireless?
Cheers,
Pat...
No trees were killed in the sending of this message.
However a large number of electrons were terribly inconvenienced.
|
|
|
|
|
the_fbi
|
| posted on 26/3/06 at 10:45 PM |
|
|
Even if due to distance you end up with only a 1Mb connection on your wireless, it'll still be fast enough for your 60k/sec.
|
|
|
greglogan
|
| posted on 26/3/06 at 11:01 PM |
|
|
I have wireless in the house. It's pretty good. If I had a big file to download tho, I would do it on the desktop as it is hard wired.
Women are meant to be loved, not understood.
|
|
|
Avoneer
|
| posted on 26/3/06 at 11:16 PM |
|
|
I'm only on (and will be on) 1Mb broadband anyway.
So basically, no matter how big the file is, I'm probably not going to loose any connection speed?
Cheers,
Pat...
No trees were killed in the sending of this message.
However a large number of electrons were terribly inconvenienced.
|
|
|
the_fbi
|
| posted on 26/3/06 at 11:46 PM |
|
|
quote:
Originally posted by Avoneer
I'm only on (and will be on) 1Mb broadband anyway.
So basically, no matter how big the file is, I'm probably not going to loose any connection speed?
If you're on 1Mb you should see speeds around 116K/sec max, presuming its coming from a good source.
Either way, as long as you are within range of your wireless, the connection speed will be at least as fast as your internet connection is 
Irrespective of file size your wireless will be as good as a wire.
|
|
|
nicksertis
|
| posted on 27/3/06 at 02:46 AM |
|
|
There are different speeds of wireless, if you get 802.11g (the letter here denotes the speed amongst other things) then that will cope with speeds
upto 54Mbps. 802.11b supports speeds of about 11Mbps and 802.11a is also 54 Mbps (but at a higher GHz - 5GHz I think)
I find that using same brand components across the board (so long as its not already built into your laptop) ensures that there are less setup issues
in the long run.
Here's a link to a smarter nerd (than I) explaining it :
http://www.microsoft.com/windowsxp/using/networking/expert/b
owman_02september03.mspx
Hope that helps,
Nick
|
|
|
chockymonster
|
| posted on 27/3/06 at 07:39 AM |
|
|
I use 802.11 superg stuff, 108MB wireless. The PC is connected by cable to the access point and the laptops use wireless.
The only time you notice that you're on wireless is copying large files between laptop and PC, browsing and downloading from the web is as fast
on the laptop as the PC.
My Bb link is 2mb, downloading from the web I get 256kb a second, the laptops can copy from the pc at a theoretical 80MB a second, in reality it works
out at around 40-50, but still a lot more than 256k.
|
|
|
DaveFJ
|
| posted on 27/3/06 at 08:47 AM |
|
|
the trick here is to remeber that your braodband connection is always quoted in 'bits per second' and your home network is rated in
'bytes per second'
there are 8 bits to the byte so....
a 1Mbps broadband link is equvalent to...
0.125 MBps !
As even the worst home networking easily exceeds this you should see no problems
HOWEVER - before comitting to wireless try to find out if your neighbours are using it. from my living room I can 'see' 6 wireless routers
and consequently the intereference is so great that wireless is totally unusable. Although there are 11 theoretical bands to use they are so close
together that they still interfere so are actually quite pointless....
I have resorted to using homeplugs which plug into your power supply. very simple and effective. they give me a consistent 84MBps and if I am using
the laptop then I almost always plug the power supply in anyway so it makes no great difference to plug in a homeplug as well. in fact they are the
dogs danglies!
Dave
"In Support of Help the Heroes" - Always
|
|
|
Vixus
|
| posted on 28/3/06 at 03:12 PM |
|
|
Wireless does not give you a significant decrease in speed as far as I can see.
{ e = mc something ~ processed meat ~ dA }
|
|
|
Gav
|
| posted on 1/4/06 at 08:31 PM |
|
|
quote:
Originally posted by DaveFJ
the trick here is to remeber that your braodband connection is always quoted in 'bits per second' and your home network is rated in
'bytes per second'
As far as im aware all networking products ive every seen have been quoted as bits per second never Bytes.
although in fairness when you downloading stuff from a web browser etc they are usually quoted in KiloBytes and MegaBytes.
ie 512kb/sec = 64KB/sec.
Also to qualify it a bit more, 100MB/sec home networks are a bit pointless because evan a decent home pc with raid0 on sata disks will only manage a
good 30MB/sec or so.
Although gigabit cards/switchs are easily available these are still only 125MB/sec and are more suited to high end file servers.
a recent test i did on a hp server with raid 5 scsi arrangement managed to push 120MB/sec.
|
|
|
the_fbi
|
| posted on 1/4/06 at 08:47 PM |
|
|
quote:
Originally posted by Gav
Also to qualify it a bit more, 100MB/sec home networks are a bit pointless because evan a decent home pc with raid0 on sata disks will only manage a
good 30MB/sec or so.
Although gigabit cards/switchs are easily available these are still only 125MB/sec and are more suited to high end file servers.
a recent test i did on a hp server with raid 5 scsi arrangement managed to push 120MB/sec.
Don't quite know where you are getting your speeds from.
A gig network won't see more than 100MB/sec of throughput, most likely around 45-50MB/sec depending on the protocols and applications being
used.
A 100Mb network would be around 10MB/sec of throughput.
An ATA100/133 drive should see an easy 70-80MB/sec read/write performance, in RAID0 even more.
Ditto SATA drives.
A nice SCSI or IDE RAID5 setup can do 800MB/sec easy.
|
|
|
Gav
|
| posted on 1/4/06 at 09:26 PM |
|
|
quote:
A gig network won't see more than 100MB/sec of throughput, most likely around 45-50MB/sec depending on the protocols and applications being
used.
a gigabit network is by definition 1000megabits per second.
1000/8 = 125megabytes per second.
this is the raw processing capability, protocol overhead is a different matter.
quote:
A 100Mb network would be around 10MB/sec of throughput.
Again 100megabit is 12.5megabyte.
quote:
An ATA100/133 drive should see an easy 70-80MB/sec read/write performance, in RAID0 even more.
Apologies 30MB/sec is a little low, ive just retested my home system which gave a 60MB/sec which is Raid0 with 2 sata disks
edit: interstingly enough i just tested my dell server which is hosted over in the US, for a single SCSI disk it managed 58MB/sec
quote:
A nice SCSI or IDE RAID5 setup can do 800MB/sec easy.
Im sorry but this is utter rubbish!, perhaps your thinking megabits not megaBYTES
For example lets take the Adaptec SCSI RAID 3410S controller. it has a maximum capacity of 640MB/sec and thats for upto 60 SCSI devices!
link for Adaptec site
here
Also take a look at the Adaptec SCSI RAID 2000S it can have up to 160MB/s per channel and 30 devices per channel.
[Edited on 1/4/06 by Gav]
[Edited on 1/4/06 by Gav]
|
|
|
the_fbi
|
| posted on 1/4/06 at 10:02 PM |
|
|
quote:
Originally posted by Gav
quote:
A gig network won't see more than 100MB/sec of throughput, most likely around 45-50MB/sec depending on the protocols and applications being
used.
a gigabit network is by definition 1000megabits per second.
1000/8 = 125megabytes per second.
this is the raw processing capability, protocol overhead is a different matter.
quote:
A 100Mb network would be around 10MB/sec of throughput.
Again 100megabit is 12.5megabyte.
quote:
An ATA100/133 drive should see an easy 70-80MB/sec read/write performance, in RAID0 even more.
Apologies 30MB/sec is a little low, ive just retested my home system which gave a 60MB/sec which is Raid0 with 2 sata disks
edit: interstingly enough i just tested my dell server which is hosted over in the US, for a single SCSI disk it managed 58MB/sec
quote:
A nice SCSI or IDE RAID5 setup can do 800MB/sec easy.
Im sorry but this is utter rubbish!, perhaps your thinking megabits not megaBYTES
For example lets take the Adaptec SCSI RAID 3410S controller. it has a maximum capacity of 640MB/sec and thats for upto 60 SCSI devices!
link for Adaptec site
here
Also take a look at the Adaptec SCSI RAID 2000S it can have up to 160MB/s per channel and 30 devices per channel.
[Edited on 1/4/06 by Gav]
[Edited on 1/4/06 by Gav]
Gav: I said I gig network won't give more than 100MB/sec of throughput. Don't just quote a simple maths calc. This isn't
throughput.
I'll look out my SCSI RAID benchmarks tomorrow for you.
Do you actually have any real world experiances of gig networks and RAID or are you just quoting based on theoretical stuff or what you've
read?
Its important people don't get incorrect information and expect 125MB/sec (as you suggest they will) from gig networks as it simply won't
happen.
Irrespective of OS, NIC or who set them up, its just not going to get close.
Must be a pretty poor Dell system to only get 58MB/sec, or heavily loaded during your test. Was that measured using hdparm?
|
|
|
the_fbi
|
| posted on 1/4/06 at 10:08 PM |
|
|
Just ran a test on a reasonably old 250GB IDE on my desktop (ATA100).
No raid, just a single disk/volume.
I'll try to lookout my old benchmarks for the SCSI's tomorrow.
|
|
|
Gav
|
| posted on 1/4/06 at 10:43 PM |
|
|
I have been working as a programmer and systems engineer for the last eight years so yes i have real work experience.
Sorry yes, i am quoting theortical performance, real world performance is quite differnet.
The dell server specs as listed by the supplier are :
1 Dell 1600SC Dual Xeon MB DAT54AMB8C2
2 Intel 2.8 GHz 533FSB P4 Xeon
4 Generic 512 MB DDR 266 ECC Reg
2 Seagate 73GB:SCSI:Ultra320 10kRPM ST373207LC
So its not a bad system at all
I ensured that apache was off while i did
a hdparm -t /dev/sdb.
Ive just redone this test to make sure nothing was causing load and the output was:
/dev/sdb1:
Timing buffered disk reads: 192 MB in 3.02 seconds = 63.58 MB/sec
so a little better than last time
Id be interested to see your SCSI benchmarks.
The image you posted for your own disk, what are the do the values mean for the X axis of that graph and the read/write, more importantly what are the
units!
id be interested to know how much data its shifting for that test, is it bigger than the disk buffer? i find it extremly hard to beleive that a
standard sata disk will provide that performance. what app is it you used to bench mark? i wouldnt mind grabbing a copy and testing my disk, i can
send you a copy of the one i use to see if their are any differences, should be quite an interesting comparason!
The program i have used for the last 5 years or so i got with a maxtor disk, it creates a 100MB file and uses that for tests.
the laptop im on now gives around 20MB/sec where as my Raid0 sata setup gives 60mb/sec. This is the same program i used on our hp servers which
resulted in around 120MB/sec so looks to me as it is a pretty consistant app, i have also run it side by side with windows performance counters and
came up with the same results.
Would you not care to comment on the scsi controller specs i presented in light of your 800MB/sec claim?
Bearing in mind the hp machines i use day in/day out are 3 disk raid5 setups with 10k disks, while being a good system is no where near the ultimate
in raid technology im sure.
I have also worked with Sun enterprise servers that had disk clusters linked via fiber optic cables, these were poo hot, however i dont have any
benchmark results for them.
edit: just for interest ive just re-run the hdparm test on my scsi server with a full load and returned 8MB/sec
[Edited on 1/4/06 by Gav]
|
|
|
the_fbi
|
| posted on 2/4/06 at 08:57 AM |
|
|
http://www.ukcartuning.co.uk/atto.zip
Thats the benchmark app.
The left side is file size used for the test.
|
|
|
the_fbi
|
| posted on 2/4/06 at 10:12 AM |
|
|
Can't find my other benchmarks at all, hopefully they are on a memory stick at work, but this is a single U320 10K drive. Not a good one either.
A nice 15K would easily give 50% better performance.
|
|
|
Gav
|
| posted on 2/4/06 at 11:23 AM |
|
|
Ive attached the test program i use
|
|
|
Gav
|
| posted on 2/4/06 at 12:20 PM |
|
|
OK ive had a play with this test app and its well compelete and utter rubbish!.
I also managed to get it to report that my disk in this laptop can read over 1000MB/sec which is cleary never going to happen from this disk!.
Looking at the sizes of files i think they are still in the cache after writting and thats why its giving misleading reports.
whlile i agree that this result is vaild if you understand whats its doing, its not vaild in the real world.
So to reiterate a real world 800MB/sec "easily" from a raid5 setup is complete and utter rubbish!
If you want a real world test i can knockup a program that writes 100MB to disk then re-reads it, then simpy count how long it took to do it, this
file is large enough not to be affected by cache results.
|
|
|
the_fbi
|
| posted on 2/4/06 at 01:53 PM |
|
|
quote:
Originally posted by Gav
OK ive had a play with this test app and its well compelete and utter rubbish!.
You better tell the manufacturers then, who claim they are "a global leader of storage connectivity and infrastructure solutions for
data-intensive computing environments."
http://www.atto.org/attoview.html
You can change the file size within atto for the test.
So the fact that atto and hdparm report results within a couple of MB/sec of each other on the same hardware, and you're saying atto is rubbish?
|
|
|
the_fbi
|
| posted on 2/4/06 at 01:59 PM |
|
|
Also, make sure you have Direct I/O ticked, this bypasses any/all software caching, else you will get stupidly high results, possibly the reason for
your GB/sec?
|
|
|
Gav
|
| posted on 2/4/06 at 05:06 PM |
|
|
ok yes you can change the file size to a whopping 1024kb, from reading the help file it will repeat the test with that transfer size to the length you
specify below it. still its going to be smaller than a modern hard disks cache (and your disk must be fairly modern given its size).
I am well aware of the options, Direct IO, it simply tells the processor to do the copying byte for byte (hence heavy cpu usage with this ticked)
rather than using UDMA which is where you tell it to copy from address space A to address space B and notify when its done.
This wont affect the disk's own memory caching ability though.
Fiddling with the settings the lowest i could get this app to report my disk was 48MB/sec which iirc is roughly what the burst speed was for this disk
and burst speeds arnt real world speeds.
Im not saying these results are wrong. just the fact that you have to understand what its doing to make sense of the results.
perhaps thats why you claimed 800MB/sec?
You still havent told me how you can possibly get a raid 5 system to 800MB/sec when adaptecs own high end specs clearly dosnt go to that sort of
bandwidth.
|
|
|
the_fbi
|
| posted on 2/4/06 at 05:14 PM |
|
|
quote:
Originally posted by Gav
You still havent told me how you can possibly get a raid 5 system to 800MB/sec when adaptecs own high end specs clearly dosnt go to that sort of
bandwidth.
Simple, it was a caching controller.
|
|
|
Gav
|
| posted on 2/4/06 at 06:37 PM |
|
|
hehe 
|
|
|
Avoneer
|
| posted on 2/4/06 at 08:23 PM |
|
|
Enough of the technical stuff - my heads hurting now.
Seen a Belkin wireless router with USB print server all in one.
So in theory I can have this in in the cupboard under the stairs connected to the NTL inlet (also in this cupboard) and can also put my printer in
there connected to the wireless router.
Then I'll need a wireless network card in my base pc (another Belkin one with an ariel thing on the back).
And then a card in my laptop when I get it.
Does that sound right for a wireless dummy?
Is there anything else I need to look for or consider?
Pat...
No trees were killed in the sending of this message.
However a large number of electrons were terribly inconvenienced.
|
|
|
%20(WinCE).JPG)
