• We’re currently investigating an issue related to the forum theme and styling that is impacting page layout and visual formatting. The problem has been identified, and we are actively working on a resolution. There is no impact to user data or functionality, this is strictly a front-end display issue. We’ll post an update once the fix has been deployed. Thanks for your patience while we get this sorted.

100 verses 1000 speed

JCROCCO

Senior member
How much of a performance boost is gigabyte anyway? running WinXP PRo P2P with NAS device and 10/100 printer servers.

Installing new equipment in new office. Used cat5E cable standard. Is paying extra for gigabyte really worth it? Usually editing files on the NAS that are 1-2 meg or even less in Autocad. Would the extra expense of a gig switch, and gig network adapters for all the computers really have a payback?
 
For small files and for print jobs, the delays in accessing the files and printer are going to pretty much overwhelm any speed advantage of Gigabit Ethernet. If you are sending gigabyte files back and forth, then, yeah, you'll see a major difference.
 
A NAS device you can afford won't take advantage of gigabyte network speeds, and printers certainly won't. Hell, most of mine are still on 10.

Gigabyte network speeds are mostly an advantage when moving big files all the time, or serious data streaming.
 
On the typical small LANs, I have the wiring there but only provide the servers with gigibit NICs and switch ports. Many of the 24 port switches out there are really 26, with 2 gigabit ports.
The servers have something other than ata100 drives, usually SCSI.
If 3 or 4 users go after a drawing or largish file at the same time, it seems to help over an all 100 network.
 
That's because the $40 Best-Buy special 10/100 switch is the bottleneck and not the server.

Please tell me you aren't running gigabit NIC's off a 10/100 switch and expecting a performance increase 🙂
 
I run gigabit at home with 3 systems and 1 switch. Do I notice a difference? Hell yes.

Typical file transfer @ 1gb : 4GBs - 2 Mins, @100mb - 12mins.

So if you really think saving 10mins is significant, then by all means.
 
Thank you 🙂

Most frustrating conversation I ever had was with a couple of Mac users that were convinced you didn't need gigabit switches if you used a new Apple G5 because Apple's handled TCP/IP differently than PC's.

 
Originally posted by: spikespiegal
Thank you 🙂

Most frustrating conversation I ever had was with a couple of Mac users that were convinced you didn't need gigabit switches if you used a new Apple G5 because Apple's handled TCP/IP differently than PC's.

Well MACs do handle TCP/IP differently than PCs. Much more efficient.

Not to throw fuel on the fire or anything.
🙂
 
Originally posted by: spikespiegal
That's because the $40 Best-Buy special 10/100 switch is the bottleneck and not the server.

Please tell me you aren't running gigabit NIC's off a 10/100 switch and expecting a performance increase 🙂

Read much?

Many of the 24 port switches out there are really 26, with 2 gigabit ports.
You can drop a copper 1000Tx adapter in a gbic slot on a high dollar name brand swtich for the same effect.

I give the gigabit ports to servers. The switch backplane can handle way more than a gig or two, and this does reduce the bottleneck into the common point, the server.
Say the clients are moving some 40Mb autocad files, which takes a handful of seconds on a 100 connection. That 's great, till three of them send or grab one from the server at the same time. The gigabit bandwidth to the server alleviates the bottleneck to the extent possible.
The rest of the offices move tiny text files, and have no use for gig-e to the desktop yet.


To answer the OP's question, yes, it is worth it to give gigabit bandwidth to fileservers. Buying a switch with a couple if gig ports is not prohibitive. From your description of the common traffic, it is not worth gearing up a 48 drop office to all gigabit, when the average transfer is so small.

On the other hand, if the office is a total of 16 computers or less, the price point of unmanaged gigabit switches is so close to that of a 10/100 switch it is not worth buying a 10/100 switch.

You don't have to spring for all new gigabit nics if you do get a gig switch, either.
The receptionist and the shipping guy with email and a couple of word docs do not need it and never will.
The server and the autocad users could.
 
Originally posted by: spidey07
Well MACs do handle TCP/IP differently than PCs. Much more efficient.

Not to throw fuel on the fire or anything.
🙂

MS has indirectly admitted this, by touting its more efficient network stack in Vista, so there's no real debate, so to speak. I've also seen some tests showing this for Linux (at least gigabit, where such issues are probably better highlighted). I expect to see a lot more discussion and testing hopefully on this when Vista is working.

I actually spent ages downloading and installing Vista CTP to see this, and was simply so disappointed by the performance and stability of it that I uninstalled it. I'll wait for the next release..
 
I've seen similar reports about IE7.
I used to try out the beta releases, but I don't have the time to do somebody else's testing anymore.
 
Originally posted by: Madwand1

MS has indirectly admitted this, by touting its more efficient network stack in Vista, so there's no real debate, so to speak. I've also seen some tests showing this for Linux (at least gigabit, where such issues are probably better highlighted). I expect to see a lot more discussion and testing hopefully on this when Vista is working.

I actually spent ages downloading and installing Vista CTP to see this, and was simply so disappointed by the performance and stability of it that I uninstalled it. I'll wait for the next release..

yeah, the MS stack has always been pretty poor. But the word is vista will be much improved and starting to follow the rules of TCP. It has been getting better and better starting with XP.
 
Originally posted by: skyking
I've seen similar reports about IE7.
I used to try out the beta releases, but I don't have the time to do somebody else's testing anymore.

I think he is referring to the refinement of the TCP/IP stack in Vista, not in IE7. IE7 with XP is still using the XP based TCP/IP stack versus the Vista stack that I believe you are referring to.
 
Ha, pretty funny. But still the way people have who have no idea about computers tell people with actual knowlege how computers work. Also the protocol is the same between mac/windows/linux/solaris/uinix/etc. So he is right about TCP/IP being handled the same on each OS (in an abstract way).

Anyway...

Bunbun
 
Originally posted by: Nocturnal
Originally posted by: skyking
I've seen similar reports about IE7.
I used to try out the beta releases, but I don't have the time to do somebody else's testing anymore.

I think he is referring to the refinement of the TCP/IP stack in Vista, not in IE7. IE7 with XP is still using the XP based TCP/IP stack versus the Vista stack that I believe you are referring to.

I was referring to general instability and dissapointment. Nothing to do with tcp/ip or anything else, it was a comment about beta testing in general.
 
It's gigabit ethernet, not gigabyte. Gigabyte ethernet would almost certainly be overkill 😛. For multiple users grabbing decently sized files (10+ MB) moving to gigabit may help out. You could always set up a couple systems on a gigabit LAN and see if there's a tangible difference that you feel is worth the added cost.
 
To get back to the OP's question with some numbers -- on file transfers, from single IDE to IDE drives, I see around 30 MB/s file transfers over (consumer) gigabit. So this would be around 3x the performance of 100 Mb/s ethernet, so well worth a marginal cost, if you do a lot of file transfers. (However, not everyone sees this sort of performance; other bottlenecks can apply; some people report only getting around 12-15 MB/s.)

Let's assume 100 Mb/s ethernet gives you 10 MB/s for convenience. What does the 3x performance gain get you? For a 10 MB file, you'd go down from 1s to 1/3s at best, and more likely, this performance difference would be swallowed up by other file access overhead, so you wouldn't be able to notice the difference for a single file. For a 90 MB file, it would go from 9s to 3s -- noticable, but easily forgettable for a single file. For a 600 MB file (e.g. a CD ISO), it would go from 60s to 20s -- here you'd easily see the difference when transferring such files.
 
Back
Top