1000 Base t

[russ Boehle]

this has come up before, but 1000 base t cards are getting pretty cheap. Has anyone tried them with A5. I was wodering if they might eliminate the need for shadowing.

Russ

[Louis Nickerson]

RE: 1000 Base t

Russ,

Let me preface this with "This is what I was taught"

I have not tried this but keep in mind...

The chain is only as strong as its weakest link ...and...

Data can only travel as fast as the slowest network component will allow.

If you have a Gigabit Network Interface Card that is connected to cabling that can only pass 100 meg then your data is only going to travel at 100 meg.

To achieve true Gigabit, ALL components must be Gigabit. This includes NICs, patch cables, jacks, patch panels, switches, routers, firewalls, and any other device or component the data will travel across.

Louis

[CraigSchumacker]

RE: 1000 Base t

Russ:

There is much more to it than replacing just the cards. All routers, switches, and hubs muxt be replaced with devices that handle 1000 mbs. In addition, the cabling must be CAT 5e... This is all where the bulk of the cost lies.

That said, my opinion is that the need for shadowing will still exist. Issues with speed are sourced more with the number of requests to the server than with the throughput of the cabling.

Of course, I would think there would be a noticeable improvement overall, especially on large networks with dozens of workstations. The more compressed the the existing datastream, the more significant the improvement.

However, my guess would be that upgrading the server would be the choice offering the most significant improvement on most systems. You might also re-evaluate your server configurations. What's it doing that it doesn't need to be doing? How redundant are your backups? Is the hard drive being defragmented on an ongoing basis?

Keep in mind that, without the shadow, the server would be required to dish up all the templates for the forms and reports in addition to whatever else it was doing before you dis-engaged the shadow. This would just further tax it's response time...

This is just my opinion - I have not installed a network with 1000 base t yet...

Craig

[russ Boehle]

RE: 1000 Base t

I'm aware of the requirements. We will be looking at redoing our offices in the next year and I'm not sure whether we could get any $ benefit from gig ethernet or not. Now we use 100 meg ethernet and A4v6 is downright snappy on it. A5v5 however is a little sluggish. I do need to upgrade the server though. (I have it but have been too busy to install it :-( )


Russ

[L. Lautenbach]

RE: 1000 Base t

Have network using10/100 Tbase.
You still want to shadow.

I also have DSl am sharing all network services on 5 computers. 1 of the 5 is in my home office 478 Ft. from router. Had a under ground telephone cable w/5 pair wires. Hooked my Network to 4/pair and terminated with Cad5. This I was told would "definetly" not work.
Well my download speeds on the office computer ave. 50-60kbs, and network is as fast or faster than ones in the main building!

[AaronBBrown]

RE: 1000 Base t

Alpha Five will run fine with a 1000baseT network. In fact, A5 doesn't care what the underlying network looks like...as long as windows works with the network, A5 will.

However, be warned that gigabit over copper is not all it's cracked up to be. You may be disappointed with the performance increase.

[M. Zijlstra]

RE: 1000 Base t

We've recently upgraded our network, and use 1000 base T for our servers and switches on every floor of the building to serve every floor with enough bandwidth.

It made everything faster in our network. Apps on the (Novell 6.5) file server open way faster, as well as data copies from server to workstation and vice versa.

But our db in A5v5 is creeping forward if to serve a user with a form with many calculated fields or a complicated sets of tables.

The database did NOT change. The workstations did NOT change. How is this possible???

See
http://msgboard.alphasoftware.com/alphaphorum/read.php3?num=11&offset=0&thread=38433&sortby=lastreply&direction=desc&id=38433
for the full message...

We have no clue...

[csda1]

RE: 1000 Base t

The reason the faster network didn't help was that the network was not the weakest link. The server speed, server disk drive or server memory was the problem (assuming all workstations saw no speed increase).

I believe that you always design your applications to use the minimum resources of your server or network, because, in the long term, you will potentially reach their maximum limit.

Good application design & techniques can make an application run very smoothly, and conversely, poor design can bring ANY NETWORK OR SERVER to it's knees.

I've seen A4V6 running 100's of workstations sharing a database and running smoothly on 10BaseT network. A4V6 is about the same as A5 running with the application shadowed. Good design made it work. 100BaseT made it fly.

Regards,

Ira

[M. Zijlstra]

RE: 1000 Base t

I understand what you are saying. And I am not saying our database is the best designwise.

But if you change one parameter (to simplify a little bit) which makes general network peformance much faster, how can it bring A5 to its knees on particular forms and browses? From being not all too fast but workable generally, to not workable at all.

Adding up to our amazement, performance only in A5 seems after the installation of the new network, to be hardly dependant on the amount of concurrent users. Before the weekend in the old building, 30 users strained the app, but it was more or less workable. 5 concurrent users had a fast app.

Now 1 or 30 users does not really matter. The forms and browses (which did NOT change over the weekend) that are now VERY slow are slow even with 1 user working on it.

As we only upgraded (and thus changed) our network, it must be something with the network and network dependant behaviour of A5. No other piece of software running on our network has the same problem, they all got faster...

Thanks for any comments so far!

Michiel Z

[jwolfe11]

RE: 1000 Base t

I've experienced this problem in the past. I will dig out old notes and make observations here.

A couple of questions:

1. What protocols are you running across this network?
2. I assume you're still using Novell. What version?
3. Do you use an accounting program or other database program in the client/server scenario? (ie Quickbooks, DacEasy,etc.)

I'll see what I can dig up as I've done alot of testing with this speed issue and A5. I do know that there was an issue with Windows NT networks and Win 95/98 workstations but this doesn't seem to be the case here.