Re: Windows 2003 VPN Speed limit??

VPN will always sync to and run at the slower "upload speed" of asyncronous
connections. Then you have to add to that the fact the VPN has a high
Protocol Overhead where some of the speed is eaten up by just making the
Protocol function.

Data *never* transfers at 100% of the line spedd on any link or any
kind,...if it did it would stop the whole network over that one physical
link untill the transfer was done, and we know that doesn't happen. So
combine the fact the you are only running at the slower upload speed of the
line, have a high protocol overhead, and the fact the data transfers don't
use the full line speed,..and you have most of your answer.

You measurement methods of how fast you get the data from the database will
always be questionable,...I don't believe there is even a real way to
measure that accuartely, so what you are seeing in your measurments may not
even be correct. About the only kind of transfer that I know I can measure
is IE downloading from a website and it is clearly stated in the Progress
Dialog box,...anything else is always going to be a questionable thing to
me. Even "timing" it won't be completely valid since the speed may not
remain consistant throughout the transfer,...at best you are only getting an
average.

Anything beyond that, I don't know what to tell you.


--
Phillip Windell [MCP, MVP, CCNA]
www.wandtv.com

The views expressed are my own (as annoying as they are), and not those of
my employer or anyone else associated with me.
-----------------------------------------------------


"Lincoln Ling" <(E-Mail Removed)> wrote in message
news:2549FD5D-E8F0-4B12-9806-(E-Mail Removed)...
> Background.
>
> I am testing a new deployment with running a small database program
> (around
> 15 MB) at the server and having clients VPN using PPTP to pull the reports
> from the database using the local clients.
> - Server backbone - Cogent 100 Mb/s fibers optical (with only 4%
> utilization on the entire network- a lot of bandwidth left)
> - Clients use DSL connection from Home (or from within the office on a
> different subnet)
>
> Problem.
> -When i schedule 10 testers to logon and pull data from the database they
> can only pull around 1.5Mb/s data?? (even with one user the upload speed
> from
> the server on the VPN clients is painfully slow only around 50kbps)
>
> We run some time and a single VPN will be able to download around 2Mb/s
> data
> from the server when downloading files. on the LAN when they VPN into the
> server they are able to pull around 40Mb/s from this server
>
> Secondly, VPN connection will sometime suddenly get disconnected (even
> from
> head office running on LAN)
>
> WHY? Routing issue?
>

To add to Phillip's comments, VPN and database operations are not a good
idea in general. Apart from the actual data transfer rates, VPN connections
are not really responsive enough for database operations. Requests can time
out because of the latency in the system. A better plan would be to use
Terminal Services, so that both database client and server are on the fast
Ethernet connection.

"Phillip Windell" <@.> wrote in message
news:eIefc%(E-Mail Removed)...
> VPN will always sync to and run at the slower "upload speed" of
> asyncronous connections. Then you have to add to that the fact the VPN has
> a high Protocol Overhead where some of the speed is eaten up by just
> making the Protocol function.
>
> Data *never* transfers at 100% of the line spedd on any link or any
> kind,...if it did it would stop the whole network over that one physical
> link untill the transfer was done, and we know that doesn't happen. So
> combine the fact the you are only running at the slower upload speed of
> the line, have a high protocol overhead, and the fact the data transfers
> don't use the full line speed,..and you have most of your answer.
>
> You measurement methods of how fast you get the data from the database
> will always be questionable,...I don't believe there is even a real way to
> measure that accuartely, so what you are seeing in your measurments may
> not even be correct. About the only kind of transfer that I know I can
> measure is IE downloading from a website and it is clearly stated in the
> Progress Dialog box,...anything else is always going to be a questionable
> thing to me. Even "timing" it won't be completely valid since the speed
> may not remain consistant throughout the transfer,...at best you are only
> getting an average.
>
> Anything beyond that, I don't know what to tell you.
>
>
> --
> Phillip Windell [MCP, MVP, CCNA]
> www.wandtv.com
>
> The views expressed are my own (as annoying as they are), and not those of
> my employer or anyone else associated with me.
> -----------------------------------------------------
>
>
> "Lincoln Ling" <(E-Mail Removed)> wrote in message
> news:2549FD5D-E8F0-4B12-9806-(E-Mail Removed)...
>> Background.
>>
>> I am testing a new deployment with running a small database program
>> (around
>> 15 MB) at the server and having clients VPN using PPTP to pull the
>> reports
>> from the database using the local clients.
>> - Server backbone - Cogent 100 Mb/s fibers optical (with only 4%
>> utilization on the entire network- a lot of bandwidth left)
>> - Clients use DSL connection from Home (or from within the office on a
>> different subnet)
>>
>> Problem.
>> -When i schedule 10 testers to logon and pull data from the database they
>> can only pull around 1.5Mb/s data?? (even with one user the upload speed
>> from
>> the server on the VPN clients is painfully slow only around 50kbps)
>>
>> We run some time and a single VPN will be able to download around 2Mb/s
>> data
>> from the server when downloading files. on the LAN when they VPN into
>> the
>> server they are able to pull around 40Mb/s from this server
>>
>> Secondly, VPN connection will sometime suddenly get disconnected (even
>> from
>> head office running on LAN)
>>
>> WHY? Routing issue?
>>
>
>