Sybase NNTP forums - End Of Life (EOL)

The NNTP forums from Sybase - forums.sybase.com - are now closed.

All new questions should be directed to the appropriate forum at the SAP Community Network (SCN).

Individual products have links to the respective forums on SCN, or you can go to SCN and search for your product in the search box (upper right corner) to find your specific developer center.

AIS doesn't use all Internet bandwidth

7 posts in Internet Server Last posting was on 2009-07-08 15:10:41.0Z
Dmitry Merkotan Posted on 2009-07-03 14:50:34.0Z
Date: Fri, 03 Jul 2009 17:50:34 +0300
From: Dmitry Merkotan <comrad@wgsoftpro.com>
User-Agent: Mozilla/5.0 (X11; U; Linux i686; en-US; rv:1.9.1b3pre) Gecko/20090513 Fedora/3.0-2.3.beta2.fc11 Thunderbird/3.0b2
MIME-Version: 1.0
Newsgroups: Advantage.Internet_Server
Subject: AIS doesn't use all Internet bandwidth
Content-Type: text/plain; charset=UTF-8; format=flowed
Content-Transfer-Encoding: 7bit
NNTP-Posting-Host: 89.162.137.146
Message-ID: <4a4e1a7f@solutions.advantagedatabase.com>
X-Trace: 3 Jul 2009 08:49:35 -0700, 89.162.137.146
Lines: 15
Path: solutions.advantagedatabase.com!solutions.advantagedatabase.com!89.162.137.146
Xref: solutions.advantagedatabase.com Advantage.Internet_Server:780
Article PK: 1121034

Hello everybody!
We try to test our program over Internet using AIS.
The outbound Internet speed from AIS site is 1Mbit/s, but it looks like
AIS doesn't use much of it, because max speed we've seen is about
160kbit/s!!! Our app is working too ssslllooowwwlllyyy on that speed.
Is there any way to tell AIS use more Internet bandwidth?


--
Dmitry Merkotan

SoftPro system administrator

Ukraine


Mark Wilkins Posted on 2009-07-06 23:28:57.0Z
From: "Mark Wilkins" <mark@no.email>
Newsgroups: Advantage.Internet_Server
References: <4a4e1a7f@solutions.advantagedatabase.com>
Subject: Re: AIS doesn't use all Internet bandwidth
Date: Mon, 6 Jul 2009 17:28:57 -0600
Lines: 48
X-Priority: 3
X-MSMail-Priority: Normal
X-Newsreader: Microsoft Outlook Express 6.00.2900.5512
X-MimeOLE: Produced By Microsoft MimeOLE V6.00.2900.5579
X-RFC2646: Format=Flowed; Response
NNTP-Posting-Host: 10.24.38.228
Message-ID: <4a52887c@solutions.advantagedatabase.com>
X-Trace: 6 Jul 2009 17:27:56 -0700, 10.24.38.228
Path: solutions.advantagedatabase.com!solutions.advantagedatabase.com!10.24.38.228
Xref: solutions.advantagedatabase.com Advantage.Internet_Server:784
Article PK: 1121038

Hi Dmitry,

The problem might be the time for a round trip for a packet on the network
as opposed to the data volume. The client makes a request of the server and
then waits for the response. If neither the request nor the response have
very much data associated with them, it will definitely be the round trip
cost that is the main issue. I'm sure there are many ways to test this, but
one that comes to mind is to run an SQL UPDATE statement in a loop and time
how long it takes for maybe 1000 iterations. Compare this time to a server
on the LAN.

For example,
create table test (i integer);
insert into test values (1);

Then run this 1000 times:
update test set i = i + 1;

After the first execution, the total data in the request and response will
each be less than 100 bytes (one packet each way). If the test over the
Internet connection is much slower, then it seems likely that the problem is
the round trip cost. If that is the case, then it maybe necessary to study
ways to reduce unnecessary traffic. Unfortunately, one fairly expensive
operation in terms of number of communication trips between client and
server is a direct table open when the table has a large number of indexes.
Often, it can be helpful (if possible) to replace those types of operations
with SQL statements (particularly in the case of updates).

Mark Wilkins
Advantage R&D

"Dmitry Merkotan" <comrad@wgsoftpro.com> wrote in message
news:4a4e1a7f@solutions.advantagedatabase.com...
> Hello everybody!
> We try to test our program over Internet using AIS.
> The outbound Internet speed from AIS site is 1Mbit/s, but it looks like
> AIS doesn't use much of it, because max speed we've seen is about
> 160kbit/s!!! Our app is working too ssslllooowwwlllyyy on that speed.
> Is there any way to tell AIS use more Internet bandwidth?
>
>
> --
> Dmitry Merkotan
> SoftPro system administrator
> Ukraine


Dmitry Merkotan Posted on 2009-07-07 13:18:45.0Z
Date: Tue, 07 Jul 2009 16:18:45 +0300
From: Dmitry Merkotan <comrad@wgsoftpro.com>
User-Agent: Mozilla/5.0 (X11; U; Linux i686; en-US; rv:1.9.1b3pre) Gecko/20090513 Fedora/3.0-2.3.beta2.fc11 Thunderbird/3.0b2
MIME-Version: 1.0
Newsgroups: Advantage.Internet_Server
Subject: Re: AIS doesn't use all Internet bandwidth
References: <4a4e1a7f@solutions.advantagedatabase.com> <4a52887c@solutions.advantagedatabase.com>
In-Reply-To: <4a52887c@solutions.advantagedatabase.com>
Content-Type: text/plain; charset=windows-1251; format=flowed
Content-Transfer-Encoding: 8bit
NNTP-Posting-Host: 89.162.137.146
Message-ID: <4a534af9@solutions.advantagedatabase.com>
X-Trace: 7 Jul 2009 07:17:45 -0700, 89.162.137.146
Lines: 48
Path: solutions.advantagedatabase.com!solutions.advantagedatabase.com!89.162.137.146
Xref: solutions.advantagedatabase.com Advantage.Internet_Server:786
Article PK: 1121040

07.07.2009 02:28, Mark Wilkins ГЇГЁГёГҐГІ:

> Hi Dmitry,
>
> The problem might be the time for a round trip for a packet on the network
> as opposed to the data volume. The client makes a request of the server and
> then waits for the response. If neither the request nor the response have
> very much data associated with them, it will definitely be the round trip
> cost that is the main issue. I'm sure there are many ways to test this, but
> one that comes to mind is to run an SQL UPDATE statement in a loop and time
> how long it takes for maybe 1000 iterations. Compare this time to a server
> on the LAN.
>
> For example,
> create table test (i integer);
> insert into test values (1);
>
> Then run this 1000 times:
> update test set i = i + 1;
>
> After the first execution, the total data in the request and response will
> each be less than 100 bytes (one packet each way). If the test over the
> Internet connection is much slower, then it seems likely that the problem is
> the round trip cost. If that is the case, then it maybe necessary to study
> ways to reduce unnecessary traffic. Unfortunately, one fairly expensive
> operation in terms of number of communication trips between client and
> server is a direct table open when the table has a large number of indexes.
> Often, it can be helpful (if possible) to replace those types of operations
> with SQL statements (particularly in the case of updates).
>

Hi, Mark!
We've done suggested test and have such results: over LAN - 0.7-0.9 sec,
over Internet - 49-50 sec. So this test over LAN is 50-60 times faster
than over Internet!!!
Our app has no direct table opening operations, we use SQL statements
everywhere.

So you think the problem is in time delays which IP packets take to
travel through Internet?

During our tests Internet channel was free of traffic on both ends.
Server and client are in different sities and hosted by different ISPs.
So is there any way to lower those time delays?

--
Dmitry Merkotan
SoftPro system administrator
Ukraine


Mark Wilkins Posted on 2009-07-07 15:28:57.0Z
From: "Mark Wilkins" <mark@no.email>
Newsgroups: Advantage.Internet_Server
References: <4a4e1a7f@solutions.advantagedatabase.com> <4a52887c@solutions.advantagedatabase.com> <4a534af9@solutions.advantagedatabase.com>
Subject: Re: AIS doesn't use all Internet bandwidth
Date: Tue, 7 Jul 2009 09:28:57 -0600
Lines: 59
X-Priority: 3
X-MSMail-Priority: Normal
X-Newsreader: Microsoft Outlook Express 6.00.2900.5512
X-MimeOLE: Produced By Microsoft MimeOLE V6.00.2900.5579
X-RFC2646: Format=Flowed; Response
NNTP-Posting-Host: 10.24.38.228
Message-ID: <4a53697c@solutions.advantagedatabase.com>
X-Trace: 7 Jul 2009 09:27:56 -0700, 10.24.38.228
Path: solutions.advantagedatabase.com!solutions.advantagedatabase.com!10.24.38.228
Xref: solutions.advantagedatabase.com Advantage.Internet_Server:788
Article PK: 1121041

Hi Dmitry,

"Dmitry Merkotan" <comrad@wgsoftpro.com> wrote in message
news:4a534af9@solutions.advantagedatabase.com...
> 07.07.2009 02:28, Mark Wilkins ïèøåò:
>
> Hi, Mark!
> We've done suggested test and have such results: over LAN - 0.7-0.9 sec,
> over Internet - 49-50 sec. So this test over LAN is 50-60 times faster
> than over Internet!!!

That is about 50 ms for a round trip then. I don't have a lot of experience
with network routing, but there might not be any major improvements
possible. I don't know if there are ways to optimize the path Internet
traffic takes. In theory, that is what the routers and various servers are
constantly trying to do already. It may be that 50 ms is as good as is
possible if there are a large number of hops between the two points. I just
ran a "ping www.yahoo.com" and it took about 50 ms per ping. A "tracert
www.yahoo.com" shows it took 18 hops. I have 7 hops from my PC and one of
our corporate servers several hundred miles away, and it takes 34 ms for
those packets. It is 30 hops and 227 ms for google (the server names shown
in the tracert indicate those packets are going through New York,
Washington, Frankfurt, Dusseldorf, and Amsterdam). I incorrectly had
assumed those packets would take a quick trip down to California rather than
an international excursion.

> Our app has no direct table opening operations, we use SQL statements
> everywhere.

That seems good to me. Using those allows for caching of table opens too.
One thing worth noting is that a live SQL statement such as "select myvalue
from thetable where id=1" will still result in an open of the actual table
by the client. But it will be cached open so that a subsequent live cursor
on that table should be fast. However, it might be faster to use "select
{static} from thetable where id=1". That would force a static cursor that
would not require any index opens. I do not know for sure if that would be
better.

>
> So you think the problem is in time delays which IP packets take to travel
> through Internet?

Yes that is what it sounds like. Running tracert might help show this.

>
> During our tests Internet channel was free of traffic on both ends.
> Server and client are in different sities and hosted by different ISPs.
> So is there any way to lower those time delays?

I do not know. You might try using TCP/IP (rather than UDP). I do not
think this would change things, but it might be worth a test. I think the
simplest way to do that is to add "use_tcp_ip=1" to the [settings] section
of the ads.ini file.


Mark Wilkins
Advantage R&D


Dmitry Merkotan Posted on 2009-07-08 07:45:37.0Z
Date: Wed, 08 Jul 2009 10:45:37 +0300
From: Dmitry Merkotan <comrad@wgsoftpro.com>
User-Agent: Mozilla/5.0 (X11; U; Linux i686; en-US; rv:1.9.1b3pre) Gecko/20090513 Fedora/3.0-2.3.beta2.fc11 Thunderbird/3.0b2
MIME-Version: 1.0
Newsgroups: Advantage.Internet_Server
Subject: Re: AIS doesn't use all Internet bandwidth
References: <4a4e1a7f@solutions.advantagedatabase.com> <4a52887c@solutions.advantagedatabase.com> <4a534af9@solutions.advantagedatabase.com> <4a53697c@solutions.advantagedatabase.com>
In-Reply-To: <4a53697c@solutions.advantagedatabase.com>
Content-Type: text/plain; charset=windows-1251; format=flowed
Content-Transfer-Encoding: 8bit
NNTP-Posting-Host: 89.162.137.146
Message-ID: <4a544e66@solutions.advantagedatabase.com>
X-Trace: 8 Jul 2009 01:44:38 -0700, 89.162.137.146
Lines: 28
Path: solutions.advantagedatabase.com!solutions.advantagedatabase.com!89.162.137.146
Xref: solutions.advantagedatabase.com Advantage.Internet_Server:790
Article PK: 1121045

Hi, Mark.

07.07.2009 18:28, Mark Wilkins ГЇГЁГёГҐГІ:

> That is about 50 ms for a round trip then. I don't have a lot of experience
> with network routing, but there might not be any major improvements
> possible. I don't know if there are ways to optimize the path Internet
> traffic takes. In theory, that is what the routers and various servers are
> constantly trying to do already. It may be that 50 ms is as good as is
> possible if there are a large number of hops between the two points. I just
> ran a "ping www.yahoo.com" and it took about 50 ms per ping. A "tracert
> www.yahoo.com" shows it took 18 hops. I have 7 hops from my PC and one of
> our corporate servers several hundred miles away, and it takes 34 ms for
> those packets. It is 30 hops and 227 ms for google (the server names shown
> in the tracert indicate those packets are going through New York,
> Washington, Frankfurt, Dusseldorf, and Amsterdam). I incorrectly had
> assumed those packets would take a quick trip down to California rather than
> an international excursion.

The distance between our server and client is ~500km, we have 8 hops to
it and ~40-45ms ping delays.
You say you have one of servers hundreds miles away and 7 hops.
It looks very close to our situation.
Could you make the test above in your environment?

--
Dmitry Merkotan
SoftPro system administrator
Ukraine


Mark Wilkins Posted on 2009-07-08 15:10:41.0Z
From: "Mark Wilkins" <mark@no.email>
Newsgroups: Advantage.Internet_Server
References: <4a4e1a7f@solutions.advantagedatabase.com> <4a52887c@solutions.advantagedatabase.com> <4a534af9@solutions.advantagedatabase.com> <4a53697c@solutions.advantagedatabase.com> <4a544e66@solutions.advantagedatabase.com>
Subject: Re: AIS doesn't use all Internet bandwidth
Date: Wed, 8 Jul 2009 09:10:41 -0600
Lines: 21
X-Priority: 3
X-MSMail-Priority: Normal
X-Newsreader: Microsoft Outlook Express 6.00.2900.5512
X-MimeOLE: Produced By Microsoft MimeOLE V6.00.2900.5579
X-RFC2646: Format=Flowed; Response
NNTP-Posting-Host: 10.24.38.228
Message-ID: <4a54b6b2@solutions.advantagedatabase.com>
X-Trace: 8 Jul 2009 09:09:38 -0700, 10.24.38.228
Path: solutions.advantagedatabase.com!solutions.advantagedatabase.com!10.24.38.228
Xref: solutions.advantagedatabase.com Advantage.Internet_Server:791
Article PK: 1121047


"Dmitry Merkotan" <comrad@wgsoftpro.com> wrote in message
news:4a544e66@solutions.advantagedatabase.com...
> Hi, Mark.
>
>
> The distance between our server and client is ~500km, we have 8 hops to it
> and ~40-45ms ping delays.
> You say you have one of servers hundreds miles away and 7 hops.
> It looks very close to our situation.
> Could you make the test above in your environment?
>

I do not have access to any of the servers in the corporate office.
However, based on the ping results, I am reasonably sure the round trip time
for ADS would also be about 33-34ms.

Mark Wilkins
Advantage R&D


Josh Neider Posted on 2009-07-07 15:23:29.0Z
From: "Josh Neider" <jneider@sybase.com>
Newsgroups: Advantage.Internet_Server
References: <4a4e1a7f@solutions.advantagedatabase.com> <4a52887c@solutions.advantagedatabase.com> <4a534af9@solutions.advantagedatabase.com>
In-Reply-To: <4a534af9@solutions.advantagedatabase.com>
Subject: Re: AIS doesn't use all Internet bandwidth
Date: Tue, 7 Jul 2009 09:23:29 -0600
Lines: 3
MIME-Version: 1.0
Content-Type: text/plain; format=flowed; charset="windows-1251"; reply-type=response
Content-Transfer-Encoding: 8bit
X-Priority: 3
X-MSMail-Priority: Normal
Importance: Normal
X-Newsreader: Microsoft Windows Live Mail 14.0.8064.206
X-MimeOLE: Produced By Microsoft MimeOLE V14.0.8064.206
NNTP-Posting-Host: 10.24.34.232
Message-ID: <4a536839@solutions.advantagedatabase.com>
X-Trace: 7 Jul 2009 09:22:33 -0700, 10.24.34.232
Path: solutions.advantagedatabase.com!solutions.advantagedatabase.com!10.24.34.232
Xref: solutions.advantagedatabase.com Advantage.Internet_Server:787
Article PK: 1121042

Dmitry,

This Knowledgebase Item may be helpful:
http://devzone.advantagedatabase.com/dz/content.aspx?Key=17&RefNo=081201-2096

The goal is to reduce the number of trips between the client and server.
Since you're already using SQL you may want to look at increasing record
caching so that more data is pulled over at once.

Regards

Josh

"Dmitry Merkotan" <comrad@wgsoftpro.com> wrote in message
news:4a534af9@solutions.advantagedatabase.com...
> 07.07.2009 02:28, Mark Wilkins ГЇГЁГёГҐГІ:
>> Hi Dmitry,
>>
>> The problem might be the time for a round trip for a packet on the
>> network
>> as opposed to the data volume. The client makes a request of the server
>> and
>> then waits for the response. If neither the request nor the response
>> have
>> very much data associated with them, it will definitely be the round trip
>> cost that is the main issue. I'm sure there are many ways to test this,
>> but
>> one that comes to mind is to run an SQL UPDATE statement in a loop and
>> time
>> how long it takes for maybe 1000 iterations. Compare this time to a
>> server
>> on the LAN.
>>
>> For example,
>> create table test (i integer);
>> insert into test values (1);
>>
>> Then run this 1000 times:
>> update test set i = i + 1;
>>
>> After the first execution, the total data in the request and response
>> will
>> each be less than 100 bytes (one packet each way). If the test over the
>> Internet connection is much slower, then it seems likely that the problem
>> is
>> the round trip cost. If that is the case, then it maybe necessary to
>> study
>> ways to reduce unnecessary traffic. Unfortunately, one fairly expensive
>> operation in terms of number of communication trips between client and
>> server is a direct table open when the table has a large number of
>> indexes.
>> Often, it can be helpful (if possible) to replace those types of
>> operations
>> with SQL statements (particularly in the case of updates).
>>
>
> Hi, Mark!
> We've done suggested test and have such results: over LAN - 0.7-0.9 sec,
> over Internet - 49-50 sec. So this test over LAN is 50-60 times faster
> than over Internet!!!
> Our app has no direct table opening operations, we use SQL statements
> everywhere.
>
> So you think the problem is in time delays which IP packets take to travel
> through Internet?
>
> During our tests Internet channel was free of traffic on both ends.
> Server and client are in different sities and hosted by different ISPs.
> So is there any way to lower those time delays?
>
> --
> Dmitry Merkotan
> SoftPro system administrator
> Ukraine