SOAP equals IIOP performace? [was Re: Why not IIOP??]

[Aleksander Slominski <as...@cs.indiana.edu>]

graham glass wrote:

> i get the impression from these discussions that people
> think that SOAP is *much* slower than IIOP. this is not true.
>
> in my own experiments, i have got SOAP/RPC running at
> almost the same speed as IIOP. when running client & server
> in the same machine, i'm getting 1.8 millisecond round trip
> times for simple SOAP/RPC messages on an average desktop.

could you tell more about it. what did you use for testing? did you get it
actually in java? or was it just sending XML text without any parsing?

> i have found that the benefits of tagged information greatly
> outweighs the tiny performance difference that IIOP or RMI gives you.

i could agree for small message sizes (latency) but what about sending bigger
data blocks (such as matrix or image)? wouldn't potential 3/2x-10x increase in
size and XML parsing hurt performance a lot?

thanks,

alek
--
Aleksander Slominski, IU, http://www.extreme.indiana.edu/~aslom
As I look afar I see neither cherry Nor tinted leaves Just a modest hut
on the coast In the dusk of Autumn nightfall-Fujiwara no Teika (1162-1241)

RE: SOAP equals IIOP performace? [was Re: Why not IIOP??]

[graham glass <gr...@mindspring.com>]

i used GLUE, my own platform for web services that i'm
about to release. it is all written in java.

the benchmarks that i mention are for the entire round trip,
including converting the call into XML, sending it via HTTP,
reading the HTTP via a servlet, parsing the XML, invoking
the call, converting the result back to XML, returning
the HTTP, parsing the result XML at the client, and then
returning the final value.

as far as XML parser performance goes, it of course all
depends on which parser you're using. i'm using Electric XML,
which is part of GLUE, and which is typically 3 to 5 times
faster than Xerces, the popular IBM XML parser.

the main point here is that it perfectly possible to
achieve good SOAP performance, and any slow performance
figures you're getting are almost certainly due to the
state of the technology you're using, not because of
anything inherent in SOAP/XML.

cheers,
graham

-----Original Message-----
From: Aleksander Slominski [mailto:aslom@cs.indiana.edu]
Sent: Wednesday, February 14, 2001 10:27 AM
To: soap-user@xml.apache.org
Subject: SOAP equals IIOP performace? [was Re: Why not IIOP??]


graham glass wrote:

> i get the impression from these discussions that people
> think that SOAP is *much* slower than IIOP. this is not true.
>
> in my own experiments, i have got SOAP/RPC running at
> almost the same speed as IIOP. when running client & server
> in the same machine, i'm getting 1.8 millisecond round trip
> times for simple SOAP/RPC messages on an average desktop.

could you tell more about it. what did you use for testing? did you get it
actually in java? or was it just sending XML text without any parsing?

> i have found that the benefits of tagged information greatly
> outweighs the tiny performance difference that IIOP or RMI gives you.

i could agree for small message sizes (latency) but what about sending
bigger
data blocks (such as matrix or image)? wouldn't potential 3/2x-10x increase
in
size and XML parsing hurt performance a lot?

thanks,

alek
--
Aleksander Slominski, IU, http://www.extreme.indiana.edu/~aslom
As I look afar I see neither cherry Nor tinted leaves Just a modest hut
on the coast In the dusk of Autumn nightfall-Fujiwara no Teika (1162-1241)

RE: SOAP equals IIOP performace? [was Re: Why not IIOP??]

[graham glass <gr...@mindspring.com>]

i used GLUE, my own platform for web services that i'm
about to release. it is all written in java.

the benchmarks that i mention are for the entire round trip,
including converting the call into XML, sending it via HTTP,
reading the HTTP via a servlet, parsing the XML, invoking
the call, converting the result back to XML, returning
the HTTP, parsing the result XML at the client, and then
returning the final value.

as far as XML parser performance goes, it of course all
depends on which parser you're using. i'm using Electric XML,
which is part of GLUE, and which is typically 3 to 5 times
faster than Xerces, the popular IBM XML parser.

the main point here is that it perfectly possible to
achieve good SOAP performance, and any slow performance
figures you're getting are almost certainly due to the
state of the technology you're using, not because of
anything inherent in SOAP/XML.

cheers,
graham

-----Original Message-----
From: Aleksander Slominski [mailto:aslom@cs.indiana.edu]
Sent: Wednesday, February 14, 2001 10:27 AM
To: soap-user@xml.apache.org
Subject: SOAP equals IIOP performace? [was Re: Why not IIOP??]


graham glass wrote:

> i get the impression from these discussions that people
> think that SOAP is *much* slower than IIOP. this is not true.
>
> in my own experiments, i have got SOAP/RPC running at
> almost the same speed as IIOP. when running client & server
> in the same machine, i'm getting 1.8 millisecond round trip
> times for simple SOAP/RPC messages on an average desktop.

could you tell more about it. what did you use for testing? did you get it
actually in java? or was it just sending XML text without any parsing?

> i have found that the benefits of tagged information greatly
> outweighs the tiny performance difference that IIOP or RMI gives you.

i could agree for small message sizes (latency) but what about sending
bigger
data blocks (such as matrix or image)? wouldn't potential 3/2x-10x increase
in
size and XML parsing hurt performance a lot?

thanks,

alek
--
Aleksander Slominski, IU, http://www.extreme.indiana.edu/~aslom
As I look afar I see neither cherry Nor tinted leaves Just a modest hut
on the coast In the dusk of Autumn nightfall-Fujiwara no Teika (1162-1241)