Menu
▾
▴
firebird-net-provider
|
From: pablosantosluac <pab...@te...> - 2007-12-27 23:41:29
|
Hi there! I'm running stress tests on Plastic with a Fb backend, and as soon as the server starts attending several clients, we've a huge performance hit! Just configuring plastic to use SQL Server solves the problem. I guess the problem could be in Fb itself, but I wonder if it could be a provider's issue. The thing is: I run a query which takes X seconds to complete. Well, two threads doing the same in parallel takes > 2xX which doesn't look like a good scalability factor. Ideas? pablo |
|
From: Dean H. <dea...@dl...> - 2007-12-28 03:36:25
|
pablosantosluac wrote: > Hi there! > > I'm running stress tests on Plastic with a Fb backend, and as soon as the > server starts attending several clients, we've a huge performance hit! Just > configuring plastic to use SQL Server solves the problem. Um, what is "plastic"? > The thing is: I run a query which takes X seconds to complete. Well, two > threads doing the same in parallel takes > 2xX which doesn't look like a > good scalability factor. Do you know the queries it is trying to execute? Can you execute them some other way (e.g. isql). That'll let you figure out if there problem is in the provider or in the server. My psychic powers tell me it's probably something to do with locking in the server, but without know what the actual queries are, it's hard to know for sure... Dean. |
|
From: Carlos <car...@gm...> - 2007-12-28 07:23:06
|
SGVsbG86CgpUaGUgdGhpbmcgaXM6IEkgcnVuIGEgcXVlcnkgd2hpY2ggdGFrZXMgWCBzZWNvbmRz IHRvIGNvbXBsZXRlLiBXZWxsLCB0d28KPiB0aHJlYWRzIGRvaW5nIHRoZSBzYW1lIGluIHBhcmFs bGVsIHRha2VzID4gMnhYIHdoaWNoIGRvZXNuJ3QgbG9vayBsaWtlIGEKPiBnb29kIHNjYWxhYmls aXR5IGZhY3Rvci4KCgpBcyBEZWFuIGhhcyBzdGF0ZWQgeWV0IHdpdGhvdXQgYSBsaXR0bGUgbW9y ZSBvZiBpbmZvcm1hdGlvbiB0aSdzIGdvaW5nIHRvIGJlCmhhcmQgdG8gaGVscCAuLi4KCldoYXQg dmVyc2lvbiBvZiBGaXJlYmlyZCBTZXJ2ZXIgYXJlIHlvdSB1c2luZyA/PwpXaGF0IHZlcnNpb24g b2YgLk5FVCBhcmUgeW91IHVzaW5nID8/CldoYXQgdmVyc2lvbiBvZiB0aGUgQURPLk5FVCBQcm92 aWRlciBhcmUgeW91IHVzaW5nID8/CgpBcmUgdGhlc2UgdGhyZWFkcyB1c2luZyB0aGVpciBvd24g Y29ubmVjdGlvbiA/PwpXaGF0IGFyZSB0aGUgcXVlcmllcyB5b3UgYXJlIHRyeWluZyB0byBydW4g Pz8KV2hhdCBpcyB0aGUgSXNvbGF0aW9uTGV2ZWwgeW91IGFyZSB1c2luZyBmb3IgdHJhbnNhY3Rp b25zID8/CgoKCgoKCi0tIApDYXJsb3MgR3V6bcOhbiDDgWx2YXJlegpWaWdvLVNwYWluCg== |
|
From: pablosantosluac <pab...@te...> - 2007-12-28 09:41:49
|
Hi Carlos,
I'll try to run again the test profiling the Firebird provider.
We're using 2.0 embedded, but same results with 2.0 (in fact the lastest =
2 available) server.
We're running with both .NET 1.1 and Mono 1.x profile.=20
We're still using the "old" 1.7.1 provider with the "charset" fix (one =
we proposed long ago and you've accepted).
We have to continue using .NET 1.1 that's why we didn't move to a newer =
provider.
Thanks!
pablo
----- Original Message -----=20
From: Carlos=20
To: For users and developers of the Firebird .NET providers=20
Sent: Friday, December 28, 2007 8:23 AM
Subject: Re: [Firebird-net-provider] Firebird scalability under heavy =
load
Hello:
The thing is: I run a query which takes X seconds to complete. Well, =
two=20
threads doing the same in parallel takes > 2xX which doesn't look =
like a
good scalability factor.
As Dean has stated yet without a little more of information ti's going =
to be hard to help ...=20
What version of Firebird Server are you using ??
What version of .NET are you using ??
What version of the ADO.NET Provider are you using ??
Are these threads using their own connection ??=20
What are the queries you are trying to run ??
What is the IsolationLevel you are using for transactions ??
=20
--=20
Carlos Guzm=C3=A1n =C3=81lvarez
Vigo-Spain=20
-------------------------------------------------------------------------=
-----
=
-------------------------------------------------------------------------=
This SF.net email is sponsored by: Microsoft
Defy all challenges. Microsoft(R) Visual Studio 2005.
http://clk.atdmt.com/MRT/go/vse0120000070mrt/direct/01/
-------------------------------------------------------------------------=
-----
_______________________________________________
Firebird-net-provider mailing list
Fir...@li...
https://lists.sourceforge.net/lists/listinfo/firebird-net-provider
|
|
From: Carlos <car...@gm...> - 2007-12-28 09:47:16
|
SGVsbG86CgpXZSBoYXZlIHRvIGNvbnRpbnVlIHVzaW5nIC5ORVQgMS4xIHRoYXQncyB3aHkgd2Ug ZGlkbid0IG1vdmUgdG8gYSBuZXdlcgo+IHByb3ZpZGVyLgo+CgoKT2ssIGl0IG1heWJlIGEgcHJv dmlkZXIgcHJvYmxlbSwgaW4gdmVyc2lvbiAyLjArIHRoZXJlIGFyZSBzZXZlcmFsIGNoYW5nZXMK aW4gdGhlIGVtYmVkZGVkIHNlcnZlciBzdXBwb3J0IHNvbWUgb2YgdGhlbSB0byBzb2x2ZXIgcGVy Zm9ybWFuY2UgcHJvYmxlbXMgKAp0aGF0IGlmIGkgcmVtZW1iZXIgd2VsbCB0aGV5IGFyZSBub3Qg YXBwbGllZCB0byAxLjcsIHNvbWUgb2YgdGhlIGNoYW5nZXMgd2FzCmNvbnRyaWJ1dHVlZCBieSBh IHByb3ZpZGVyIHVzZXIgdGltZSBhZ28gKQoKCgoKCi0tIApDYXJsb3MgR3V6bcOhbiDDgWx2YXJl egpWaWdvLVNwYWluCg== |
|
From: pablosantosluac <pab...@te...> - 2007-12-28 10:06:22
|
Ok, so, you mean I should try with the new 2.0+ provider? I mean, we are =
having the same problems with the "server" (not the embedded one).
Our intention is to continue using Firebird as our primary target =
database, so we'll really like to outperform other SCM systems using Fb =
as we do using SQL Server.
pablo
----- Original Message -----=20
From: Carlos=20
To: For users and developers of the Firebird .NET providers=20
Sent: Friday, December 28, 2007 10:47 AM
Subject: Re: [Firebird-net-provider] Firebird scalability under heavy =
load
Hello:
We have to continue using .NET 1.1 that's why we didn't move to a =
newer provider.
Ok, it maybe a provider problem, in version 2.0+ there are several =
changes in the embedded server support some of them to solver =
performance problems ( that if i remember well they are not applied to =
1.7, some of the changes was contributued by a provider user time ago )
=20
--=20
Carlos Guzm=C3=A1n =C3=81lvarez
Vigo-Spain=20
-------------------------------------------------------------------------=
-----
=
-------------------------------------------------------------------------=
This SF.net email is sponsored by: Microsoft
Defy all challenges. Microsoft(R) Visual Studio 2005.
http://clk.atdmt.com/MRT/go/vse0120000070mrt/direct/01/
-------------------------------------------------------------------------=
-----
_______________________________________________
Firebird-net-provider mailing list
Fir...@li...
https://lists.sourceforge.net/lists/listinfo/firebird-net-provider
|
|
From: Carlos <car...@gm...> - 2007-12-28 10:16:09
|
SGVsbG86CgoKIE9rLCBzbywgeW91IG1lYW4gSSBzaG91bGQgdHJ5IHdpdGggdGhlIG5ldyAyLjAr IHByb3ZpZGVyPyBJIG1lYW4sIHdlIGFyZQo+IGhhdmluZyB0aGUgc2FtZSBwcm9ibGVtcyB3aXRo IHRoZSAic2VydmVyIiAobm90IHRoZQo+CgplbWJlZGRlZCBvbmUpLgo+CgoKV2VsbCBubywgaXQg bWF5YmUgYSBwcm9ibGVtIG9uIGl0LCBidXQgd2UgbWF5IHRyeSB0byBjaGVjayB3aGVyZSBpcyB0 aGUKcHJvYmxlbSAoIHRoZSAyLjArIGhhcyBwZXJmb3JtYW5jZSBjaGFuZ2VzIG9uIG90aGVyIGFy ZWFzIGFzIHdlbGwsIGJ1dAphbnl3YXksIHdlIHNob3VsZCBjaGVjayBmaXJzdCB3aGVyZSBpcyB0 aGUgcmVhbCBwcm9ibGVtICkKCgoKLS0gCkNhcmxvcyBHdXptw6FuIMOBbHZhcmV6ClZpZ28tU3Bh aW4K |
|
From: pablosantosluac <pab...@te...> - 2008-01-13 15:03:43
|
Hi Carlos,
I've just run our test again using Firebird backend. The table below =
shows the profiler results for a test with a server (embedded) working =
with 20 simultaneous clients, so there's a lot of concurrency.
Routine Name Module Name % Time % with Children Time Time with =
Children Shared Time Hit Count=20
FesStatement::Fetch FirebirdSql.Data.Firebird.dll 18,57% 3,55% =
2679,05 7806,72 34,32 1240193=20
FesStatement::Free FirebirdSql.Data.Firebird.dll 8,40% 0,56% =
1211,19 1225,23 98,85 660413=20
XsqldaMarshaler::MarshalManagedToNative =
FirebirdSql.Data.Firebird.dll 6,66% 0,80% 960,1 1749,81 54,87 1617581=20
XsqldaMarshaler::MarshalNativeToManaged =
FirebirdSql.Data.Firebird.dll 6,51% 1,39% 939 3059,23 30,69 1563064=20
GlobalizationHelper::CultureAwareCompare =
FirebirdSql.Data.Firebird.dll 3,74% 0,25% 539,87 539,87 100 30454745=20
XsqldaMarshaler::GetString FirebirdSql.Data.Firebird.dll 3,16% =
0,31% 455,28 684,84 66,48 33472568=20
XsqldaMarshaler::CleanUpNativeData FirebirdSql.Data.Firebird.dll =
3,11% 0,24% 448,06 528,05 84,85 2883890=20
FesStatement::GetSqlInfo FirebirdSql.Data.Firebird.dll 2,34% 0,16% =
337,8 343,18 98,43 280302=20
DbField::GetDbDataType FirebirdSql.Data.Firebird.dll 2,22% 0,22% =
319,82 490,04 65,26 39600463=20
FesStatement::Execute FirebirdSql.Data.Firebird.dll 1,92% 0,16% =
277,65 351,86 78,91 660413=20
FesTransaction::Commit FirebirdSql.Data.Firebird.dll 1,89% 0,12% =
272,31 273,31 99,64 13028=20
XsqldaMarshaler::ComputeLength FirebirdSql.Data.Firebird.dll 1,88% =
0,12% 271,28 271,28 100 35290780=20
Descriptor::get_Item FirebirdSql.Data.Firebird.dll 1,88% 0,12% =
270,46 270,46 100 221582569=20
FbDataReader::GetOrdinal FirebirdSql.Data.Firebird.dll 1,83% 0,42% =
263,55 919,09 28,67 6088214=20
FesStatement::Prepare FirebirdSql.Data.Firebird.dll 1,58% 0,59% =
228,24 1291,67 17,67 197368=20
XsqldaMarshaler::MarshalManagedToNative =
FirebirdSql.Data.Firebird.dll 1,50% 0,13% 217,06 295,26 73,52 1617581=20
Ok, the first two methods (Fetch & Free) eat about 26% of the test time =
and they're basically spending their times at FbClient.isc_dsql_fetch =
and FbClient.isc_dsql_free_statement.
It looks like Marshaling and Unmarshaling data is taken a long time, I'd =
say even longer than the "internal" calls themselves.
On the other hand, in my test, the embedded server is slower than the =
"external" one: 444s to complete the embedded vs 291 the "external". =
Just to give some data to compare: using SQL Server the same test =
(exactly the same code base) requires 198 seconds. The interesting thing =
is that the test with only one client (instead of 20) is faster with the =
Fb backend than the SQL Server one. Maybe it is not an issue with the =
provider but firebird itself, although the time taken to marshal and =
unmarshal looks like a possible "culprit".
But, running exactly the same test against an external server I get the =
following:
Routine Name Module Name % Time % with Children Time Time with =
Children Shared Time Hit Count=20
XdrStream::Read FirebirdSql.Data.Firebird.dll 37,98% 2,28% 3573,73 =
3690,01 96,85 31677462=20
FbDataReader::GetOrdinal FirebirdSql.Data.Firebird.dll 3,38% 0,41% =
318,18 655,04 48,57 6072788=20
XdrStream::Write FirebirdSql.Data.Firebird.dll 2,27% 0,23% 213,13 =
375,04 56,83 12712555=20
DbField::GetDbDataType FirebirdSql.Data.Firebird.dll 2,05% 0,18% =
192,56 290,64 66,25 15044218=20
XdrStream::Flush FirebirdSql.Data.Firebird.dll 1,98% 0,12% 186,07 =
190,78 97,53 2740818=20
GlobalizationHelper::CultureAwareCompare =
FirebirdSql.Data.Firebird.dll 1,94% 0,11% 182,78 182,78 100 30456311=20
I don't know whether there is any possible optimization at the =
provider's level.
Thanks,
pablo
----- Original Message -----=20
From: Carlos=20
To: For users and developers of the Firebird .NET providers=20
Sent: Friday, December 28, 2007 11:16 AM
Subject: Re: [Firebird-net-provider] Firebird scalability under heavy =
load
Hello:
Ok, so, you mean I should try with the new 2.0+ provider? I mean, we =
are having the same problems with the "server" (not the=20
embedded one).
Well no, it maybe a problem on it, but we may try to check where is =
the problem ( the 2.0+ has performance changes on other areas as well, =
but anyway, we should check first where is the real problem )=20
=20
--=20
Carlos Guzm=C3=A1n =C3=81lvarez
Vigo-Spain=20
-------------------------------------------------------------------------=
-----
=
-------------------------------------------------------------------------=
This SF.net email is sponsored by: Microsoft
Defy all challenges. Microsoft(R) Visual Studio 2005.
http://clk.atdmt.com/MRT/go/vse0120000070mrt/direct/01/
-------------------------------------------------------------------------=
-----
_______________________________________________
Firebird-net-provider mailing list
Fir...@li...
https://lists.sourceforge.net/lists/listinfo/firebird-net-provider
|
|
From: Jiri C. <di...@ci...> - 2008-01-13 21:11:03
|
On 1/13/08, pablosantosluac <pab...@te...> wrote:
> On the other hand, in my test, the embedded server is slower than the "external" one: 444s to complete the embedded vs 291 the "external". Just to give some data to compare: using SQL Server the same test (exactly the same code base) requires 198 seconds. The interesting thing is that the test with only one client (instead of 20) is faster with the Fb backend than the SQL Server one. Maybe it is not an issue with the provider but firebird itself, although the time taken to marshal and unmarshal looks like a possible "culprit".
Well the "external" is native you can save some time spent in embedded
to pass data to/from dll library.
--
Jiri {x2} Cincura (Microsoft Student Partner)
http://blog.vyvojar.cz/jirka/ | http://www.ID3renamer.com
|
|
From: Carlos <car...@gm...> - 2008-01-14 13:35:15
|
SGVsbG86CgoKPiAgIE9rLCB0aGUgZmlyc3QgdHdvIG1ldGhvZHMgKEZldGNoICYgRnJlZSkgZWF0 IGFib3V0IDI2JSBvZiB0aGUgdGVzdCB0aW1lCj4gYW5kIHRoZXkncmUgYmFzaWNhbGx5IHNwZW5k aW5nIHRoZWlyIHRpbWVzIGF0IEZiQ2xpZW50LmlzY19kc3FsX2ZldGNoIGFuZAo+IEZiQ2xpZW50 LmlzY19kc3FsX2ZyZWVfc3RhdGVtZW50Lgo+CgpNYXliZSB0aGUgcHJvYmxlbSBjb3VsZCBiZSBp biB0aGUgZmFjdCB0aGUgZmV0Y2hlcyBhcmUgZG9uZSAgd2l0aCBvbmx5IG9uZQpyZWNvcmQgaW4g ZWFjaCB0aW1lLCB3aGlsZSB3aGVuIGRvbmUgYWdhaW5zdCB0aGUgc3RhbmRhbG9uZSBzZXJ2ZXIs IHRoZXkgYXJlCmRvbmUgaW4gYmF0Y2hlcyAoIG9mIDIwMCBwZXIgZGVmYXVsdCApLCBoYXZpbmcg aW4gbWluZCB0aGF0IHRoZSBlbWJiZWRlZApzZXJ2ZXIgc3VwcG9ydCBpcyBkb25lIHVzaW5nIFBJ bnZva2UgYW5kIHRoYXQgUEludm9rZSBjYWxscyBoYXZlIGEKcGVyZm9ybWFuY2UgcGVuYWx0eSwg dGhhdCBjYW4gYmUgcmVhc29uIG9mIHRoZSBwcm9ibGVtICggb3Igb25lIG9mIHRoZW0gOykKKS4K CkltcGxlbWVudGluZyB0aGUgZW1iZWRkZWQgc2VydmVyIHN1cHBvcnQgdXNpbmcgQysrIHdpbGwg cHJvYmFibHkgZml4IHRoZXNlCnBlcmZvcm1hbmNlIHByb2JsZW1zIC4uLiBidXQgaXQgd2lsbCBi ZSBhIGhhcmQgdGFzayBpbiBteSBvcGluaW9uLgoKSXQgbG9va3MgbGlrZSBNYXJzaGFsaW5nIGFu ZCBVbm1hcnNoYWxpbmcgZGF0YSBpcyB0YWtlbiBhIGxvbmcgdGltZSwgSSdkIHNheQo+IGV2ZW4g bG9uZ2VyIHRoYW4gdGhlICJpbnRlcm5hbCIgY2FsbHMgdGhlbXNlbHZlcy4KPgoKCi0tIApDYXJs b3MgR3V6bcOhbiDDgWx2YXJlegpWaWdvLVNwYWluCg== |
|
From: Steve M. <ste...@si...> - 2008-01-15 18:24:07
|
Hi Pablo, Thank you for publishing your numbers. > On the other hand, in my test, the embedded server is slower than the > "external" one: 444s to complete the embedded vs 291 the "external". > Just to give some data to compare: using SQL Server the same test > (exactly the same code base) requires 198 seconds. The interesting > thing is that the test with only one client (instead of 20) is faster > with the Fb backend than the SQL Server one. Maybe it is not an issue > with the provider but firebird itself, although the time taken to > marshal and unmarshal looks like a possible "culprit". Were you using SQL Server 2005? Which version were you using? I'm interested in comparing Firebird against SQL Server Express. Steve |
|
From: pablosantosluac <pab...@te...> - 2008-01-15 20:17:00
|
Hi, Well, needless to say I'm pretty happy with Firebird, which I believe is a great db. In fact up to 10 concurrent users (I'm talking about our system, it will obviously vary with others) Fb is better than SQL Server. Well, I tried both with SQLServer 2005 (a developers' release) and Express and they both give me the same numbers! I tried both with the same hardware/sw configuration and the same dedicated client machines (about 40 boxes running plastic clients). pablo ----- Original Message ----- From: "Steve Miller" <ste...@si...> To: "For users and developers of the Firebird .NET providers" <fir...@li...> Sent: Tuesday, January 15, 2008 7:24 PM Subject: Re: [Firebird-net-provider] Firebird scalability under heavy load > Hi Pablo, > > Thank you for publishing your numbers. > >> On the other hand, in my test, the embedded server is slower than the >> "external" one: 444s to complete the embedded vs 291 the "external". >> Just to give some data to compare: using SQL Server the same test >> (exactly the same code base) requires 198 seconds. The interesting >> thing is that the test with only one client (instead of 20) is faster >> with the Fb backend than the SQL Server one. Maybe it is not an issue >> with the provider but firebird itself, although the time taken to >> marshal and unmarshal looks like a possible "culprit". > > > Were you using SQL Server 2005? Which version were you using? I'm > interested in comparing Firebird against SQL Server Express. > > Steve > > ------------------------------------------------------------------------- > This SF.net email is sponsored by: Microsoft > Defy all challenges. Microsoft(R) Visual Studio 2008. > http://clk.atdmt.com/MRT/go/vse0120000070mrt/direct/01/ > _______________________________________________ > Firebird-net-provider mailing list > Fir...@li... > https://lists.sourceforge.net/lists/listinfo/firebird-net-provider |
|
From: Steve M. <ste...@si...> - 2008-01-16 16:55:00
|
> Well, needless to say I'm pretty happy with Firebird, which I believe is a > great db. In fact up to 10 concurrent users (I'm talking about our system, > it will obviously vary with others) Fb is better than SQL Server. > > > Well, I tried both with SQLServer 2005 (a developers' release) and Express > and they both give me the same numbers! I tried both with the same > hardware/sw configuration and the same dedicated client machines (about 40 > boxes running plastic clients). > That's very interesting. The Express version of SQL Server is limited to 1 processor and 1 GB of memory. Maybe I'm not awake yet, but I would have expected the opposite results than what you show. I expected that as more users connected to SQL Server, it would slow down, from being too limited, and Firebird would show more speed. Steve |
|
From: pablosantosluac <pab...@te...> - 2008-01-20 19:31:03
|
well, you know, it depends on the test, but I can tell you that I'm getting much better results out of SQL Server right now, but I'm working on this, my intention is having a fb backend much faster than anything else... and yes, I didn't see any different from Express or no-express... Are you sure it only uses one processor? ----- Original Message ----- From: "Steve Miller" <ste...@si...> To: "For users and developers of the Firebird .NET providers" <fir...@li...> Sent: Wednesday, January 16, 2008 5:54 PM Subject: Re: [Firebird-net-provider] Firebird scalability under heavy load > > >> Well, needless to say I'm pretty happy with Firebird, which I believe is >> a >> great db. In fact up to 10 concurrent users (I'm talking about our >> system, >> it will obviously vary with others) Fb is better than SQL Server. >> >> >> Well, I tried both with SQLServer 2005 (a developers' release) and >> Express >> and they both give me the same numbers! I tried both with the same >> hardware/sw configuration and the same dedicated client machines (about >> 40 >> boxes running plastic clients). >> > > That's very interesting. The Express version of SQL Server is limited to > 1 processor and 1 GB of memory. Maybe I'm not awake yet, but I would > have expected the opposite results than what you show. I expected that > as more users connected to SQL Server, it would slow down, from being > too limited, and Firebird would show more speed. > > > Steve > > ------------------------------------------------------------------------- > This SF.net email is sponsored by: Microsoft > Defy all challenges. Microsoft(R) Visual Studio 2008. > http://clk.atdmt.com/MRT/go/vse0120000070mrt/direct/01/ > _______________________________________________ > Firebird-net-provider mailing list > Fir...@li... > https://lists.sourceforge.net/lists/listinfo/firebird-net-provider |
|
From: Steve M. <ste...@si...> - 2008-01-21 15:05:14
|
> well, you know, it depends on the test, but I can tell you that I'm getting > much better results out of SQL Server right now, but I'm working on this, my > intention is having a fb backend much faster than anything else... and yes, > I didn't see any different from Express or no-express... Are you sure it > only uses one processor? > According to Microsoft's documentation, Express supports "up to 1 CPU". http://download.microsoft.com/download/4/2/c/42c6af76-a28c-4560-8269-e8fe906ea675/SQLExpressdatasheet.pdf Steve |
|
From: Dean H. <dea...@dl...> - 2008-01-21 21:13:27
|
Steve Miller wrote: >> well, you know, it depends on the test, but I can tell you that I'm getting >> much better results out of SQL Server right now, but I'm working on this, my >> intention is having a fb backend much faster than anything else... and yes, >> I didn't see any different from Express or no-express... Are you sure it >> only uses one processor? >> > > According to Microsoft's documentation, Express supports "up to 1 CPU". > http://download.microsoft.com/download/4/2/c/42c6af76-a28c-4560-8269-e8fe906ea675/SQLExpressdatasheet.pdf Of course, one CPU doesn't mean 1-connection-at-a-time... My guess is you'd still need a fair number of simultaneous connections before one CPU became the bottleneck. Don't forget also that if you're running Firebird in Superserver mode, it'll only make use of one CPU as well. Still, comparing database performance *in general* can be tricky. Some databases are better at some workloads than others, and vice versa. All you can do is compare them on the *particular* load your application creates. That does not (necessarily) mean your results are valid under ALL workloads. Having said that, there's certainly no harm in doing the comparison. Dean. |
|
From: pablosantosluac <pab...@te...> - 2007-12-28 09:36:19
|
Hi there, Plastic is www.plasticscm.com. I didn't try with isql, I would need to run several queries in parallel to actually know whether the problem is the provider or not. What I wanted to know is whether you've profiled it under concurrent access or not. Locking should be the most likely one to blame here, but you never know until you try it... pablo ----- Original Message ----- From: "Dean Harding" <dea...@dl...> To: "For users and developers of the Firebird .NET providers" <fir...@li...> Sent: Friday, December 28, 2007 4:36 AM Subject: Re: [Firebird-net-provider] Firebird scalability under heavy load > pablosantosluac wrote: >> Hi there! >> >> I'm running stress tests on Plastic with a Fb backend, and as soon as the >> server starts attending several clients, we've a huge performance hit! >> Just >> configuring plastic to use SQL Server solves the problem. > > Um, what is "plastic"? > >> The thing is: I run a query which takes X seconds to complete. Well, two >> threads doing the same in parallel takes > 2xX which doesn't look like a >> good scalability factor. > > Do you know the queries it is trying to execute? Can you execute them > some other way (e.g. isql). That'll let you figure out if there problem > is in the provider or in the server. > > My psychic powers tell me it's probably something to do with locking in > the server, but without know what the actual queries are, it's hard to > know for sure... > > Dean. > > > ------------------------------------------------------------------------- > This SF.net email is sponsored by: Microsoft > Defy all challenges. Microsoft(R) Visual Studio 2005. > http://clk.atdmt.com/MRT/go/vse0120000070mrt/direct/01/ > _______________________________________________ > Firebird-net-provider mailing list > Fir...@li... > https://lists.sourceforge.net/lists/listinfo/firebird-net-provider |
|
From: pablosantosluac <pab...@te...> - 2008-01-14 13:19:26
|
Hi,
Is there a place where I can find a performance benchmark comparing Firebird
with other dbs??
pablo
----- Original Message -----
From: "Jiri Cincura" <di...@ci...>
To: "For users and developers of the Firebird .NET providers"
<fir...@li...>
Sent: Sunday, January 13, 2008 10:11 PM
Subject: Re: [Firebird-net-provider] Firebird scalability under heavy load
> On 1/13/08, pablosantosluac <pab...@te...> wrote:
>> On the other hand, in my test, the embedded server is slower than the
>> "external" one: 444s to complete the embedded vs 291 the "external".
>> Just to give some data to compare: using SQL Server the same test
>> (exactly the same code base) requires 198 seconds. The interesting thing
>> is that the test with only one client (instead of 20) is faster with the
>> Fb backend than the SQL Server one. Maybe it is not an issue with the
>> provider but firebird itself, although the time taken to marshal and
>> unmarshal looks like a possible "culprit".
>
> Well the "external" is native you can save some time spent in embedded
> to pass data to/from dll library.
>
> --
> Jiri {x2} Cincura (Microsoft Student Partner)
> http://blog.vyvojar.cz/jirka/ | http://www.ID3renamer.com
>
> -------------------------------------------------------------------------
> Check out the new SourceForge.net Marketplace.
> It's the best place to buy or sell services for
> just about anything Open Source.
> http://ad.doubleclick.net/clk;164216239;13503038;w?http://sf.net/marketplace
> _______________________________________________
> Firebird-net-provider mailing list
> Fir...@li...
> https://lists.sourceforge.net/lists/listinfo/firebird-net-provider
|
|
From: Jiri C. <di...@ci...> - 2008-01-14 13:27:22
|
On 1/14/08, pablosantosluac <pab...@te...> wrote:
> Hi,
>
> Is there a place where I can find a performance benchmark comparing Firebird
> with other dbs??
Probably yes. But better to ask in support list.
--
Jiri {x2} Cincura (Microsoft Student Partner)
http://blog.vyvojar.cz/jirka/ | http://www.ID3renamer.com
|
×
Want the latest updates on software, tech news, and AI?
Get latest updates about software, tech news, and AI from SourceForge directly in your inbox once a month.
Thanks for helping keep SourceForge clean.
X