I think EF codefirst is the issue.
Could you explain this a little bit more?
2.0 speed/performance concerns
- 83
- 496
- 7/11/2011
- United States
fkierzek wrote:
I'm with you fkierzek.
I'm new to nopCommerce, just started looking at nopCommerce about a week ago since the port to MVC. I do like the product. But if people want this to be a commercial product you cannot count on the retail users have the latest and fastest computers. But we're not dealing with that, in terms of scalability, it is good to reduce unecessary cycles.
Sounds like a problem with the effeciency in the algorithm O(n^2). New York mentioned he saw a loop within a loop, the time in which the user have to wait becomes exponential as the number of items goes up.
And the whole "I am a cool guy cause I have a cool laptop" thing just make me laugh. I am modifying NopCommere on a laptop with 1/2 of that the horsepower without any problems. All right - fun's over. Back to business.
One of the biggest problems I've had with previous NopCommerce installations was performance when lots o products are loaded. The bottleneck seemed to be EF itself, as I had optimised SQL extensively. A reasonable way to speed things up for me was to cut on numer of items shown on page down to about 10. Pretty much catalog page load time was linearly proportional to numer of items shown. I would be interested to hear if someone noticed that to be the case with Nop 2.0.
One of the biggest problems I've had with previous NopCommerce installations was performance when lots o products are loaded. The bottleneck seemed to be EF itself, as I had optimised SQL extensively. A reasonable way to speed things up for me was to cut on numer of items shown on page down to about 10. Pretty much catalog page load time was linearly proportional to numer of items shown. I would be interested to hear if someone noticed that to be the case with Nop 2.0.
I'm with you fkierzek.
I'm new to nopCommerce, just started looking at nopCommerce about a week ago since the port to MVC. I do like the product. But if people want this to be a commercial product you cannot count on the retail users have the latest and fastest computers. But we're not dealing with that, in terms of scalability, it is good to reduce unecessary cycles.
Sounds like a problem with the effeciency in the algorithm O(n^2). New York mentioned he saw a loop within a loop, the time in which the user have to wait becomes exponential as the number of items goes up.
- 83
- 496
- 7/11/2011
- United States
fkierzek wrote:
Thanks fkierzek. People, keep in mind this is beta test, and the purpose of beta testing is to weed out problems. We're glad to help out.
32/64 - I didn't see any difference when I switched when reinstalling Windows 7.
Computer speed these days isn't a problem. O^2 is and always will.
Filip
Computer speed these days isn't a problem. O^2 is and always will.
Filip
Thanks fkierzek. People, keep in mind this is beta test, and the purpose of beta testing is to weed out problems. We're glad to help out.
- 28
- 178
- 9/10/2009
- United States
theonlylawislove wrote:
It is not the issue, I been using EF code first since CTP 4 and it isn't the issue... trust me...
I have done extensive tests with Nop 2.0 and it performs quite well on IIS 7.5
my page loads are less than 1.5 ms... So the only conclusion is that if people are experiencing performance issues than perhapse they need a better box.
I think EF codefirst is the issue.
It is not the issue, I been using EF code first since CTP 4 and it isn't the issue... trust me...
I have done extensive tests with Nop 2.0 and it performs quite well on IIS 7.5
my page loads are less than 1.5 ms... So the only conclusion is that if people are experiencing performance issues than perhapse they need a better box.
MVP
- 157
- 1753
- 9/25/2009
- United States
dswatik wrote:
I've looked with dotProfiler. The queries are taking alot longer than they should, considering my db is on the same machine. The problem is with EF's codefirst query caching. It's almost non existant. EF codefirst also has no support for compiled queries, which means that the expression tree is traversed more than it should.
I think EF codefirst is the issue.
It is not the issue, I been using EF code first since CTP 4 and it isn't the issue... trust me...
I have done extensive tests with Nop 2.0 and it performs quite well on IIS 7.5
my page loads are less than 1.5 ms... So the only conclusion is that if people are experiencing performance issues than perhapse they need a better box.
It is not the issue, I been using EF code first since CTP 4 and it isn't the issue... trust me...
I have done extensive tests with Nop 2.0 and it performs quite well on IIS 7.5
my page loads are less than 1.5 ms... So the only conclusion is that if people are experiencing performance issues than perhapse they need a better box.
I've looked with dotProfiler. The queries are taking alot longer than they should, considering my db is on the same machine. The problem is with EF's codefirst query caching. It's almost non existant. EF codefirst also has no support for compiled queries, which means that the expression tree is traversed more than it should.
- 83
- 496
- 7/11/2011
- United States
I'm changing my servers hardware around and trying to understand the performance issue better. I've upgrade my SQL Server to run on a 64-bit Windows 2008 R2 to see if this help. I've also stripped 2 1Tbytes drives together to house the nopCommerce website. The performance monitor on either machine does not show any cpu or drive or network bottlenecks.
Currently when I repeatedly click on the Category links I get between say .7 sec to about 1.5 sec. The issue here is not that it takes 1.5 seconds to get the page back, the issue at least for me is that there is a ~.7 to 1.5 seconds of inactivity before something happens, and this is on a local LAN. I'm not sure if this is the way MVC processes the request - that MVC returns the page to the client browswe only after it has rendered the complete page request. I need to look at this closer.
Also I don't know if anyone sees this: if I don't hit the home page for a long time, say I go away for 1 hr, and return to the home page. Sometimes it 60 seconds to bring up the home page, then it fine subsequent visits. I'm using internet explorer 9 and Windows Vista 32-bit.
Funny thing is when I try this on my XP laptop and IE8, this cut down to 30 seconds. Go figure? I have not tried it with my Windows 7 64-bit machine yet.
Currently when I repeatedly click on the Category links I get between say .7 sec to about 1.5 sec. The issue here is not that it takes 1.5 seconds to get the page back, the issue at least for me is that there is a ~.7 to 1.5 seconds of inactivity before something happens, and this is on a local LAN. I'm not sure if this is the way MVC processes the request - that MVC returns the page to the client browswe only after it has rendered the complete page request. I need to look at this closer.
Also I don't know if anyone sees this: if I don't hit the home page for a long time, say I go away for 1 hr, and return to the home page. Sometimes it 60 seconds to bring up the home page, then it fine subsequent visits. I'm using internet explorer 9 and Windows Vista 32-bit.
Funny thing is when I try this on my XP laptop and IE8, this cut down to 30 seconds. Go figure? I have not tried it with my Windows 7 64-bit machine yet.
- 83
- 496
- 7/11/2011
- United States
dswatik wrote:
Can you tell us about your hardware and software configuration?
I think EF codefirst is the issue.
It is not the issue, I been using EF code first since CTP 4 and it isn't the issue... trust me...
I have done extensive tests with Nop 2.0 and it performs quite well on IIS 7.5
my page loads are less than 1.5 ms... So the only conclusion is that if people are experiencing performance issues than perhapse they need a better box.
It is not the issue, I been using EF code first since CTP 4 and it isn't the issue... trust me...
I have done extensive tests with Nop 2.0 and it performs quite well on IIS 7.5
my page loads are less than 1.5 ms... So the only conclusion is that if people are experiencing performance issues than perhapse they need a better box.
Can you tell us about your hardware and software configuration?
- 217
- 2565
- 9/20/2010
- Poland
MarioRosario wrote:
Also I don't know if anyone sees this: if I don't hit the home page for a long time, say I go away for 1 hr, and return to the home page. Sometimes it 60 seconds to bring up the home page, then it fine subsequent visits. I'm using internet explorer 9 and Windows Vista 32-bit.
Funny thing is when I try this on my XP laptop and IE8, this cut down to 30 seconds. Go figure? I have not tried it with my Windows 7 64-bit machine yet.
This likely due to app pool shutting down. The default idle timeout is 20 minutes. If you go away from your computer for > 20 minutes not only the home page, but any other page will take much longer to load.
Also I don't know if anyone sees this: if I don't hit the home page for a long time, say I go away for 1 hr, and return to the home page. Sometimes it 60 seconds to bring up the home page, then it fine subsequent visits. I'm using internet explorer 9 and Windows Vista 32-bit.
Funny thing is when I try this on my XP laptop and IE8, this cut down to 30 seconds. Go figure? I have not tried it with my Windows 7 64-bit machine yet.
This likely due to app pool shutting down. The default idle timeout is 20 minutes. If you go away from your computer for > 20 minutes not only the home page, but any other page will take much longer to load.