Groups | Search | Server Info | Keyboard shortcuts | Login | Register [http] [https] [nntp] [nntps]
Groups > comp.lang.java.programmer > #13352 > unrolled thread
| Started by | Daniel Pitts <newsgroup.nospam@virtualinfinity.net> |
|---|---|
| First post | 2012-04-03 11:19 -0700 |
| Last post | 2012-04-07 10:50 -0700 |
| Articles | 7 — 4 participants |
Back to article view | Back to comp.lang.java.programmer
What load testing solutions do you use? Daniel Pitts <newsgroup.nospam@virtualinfinity.net> - 2012-04-03 11:19 -0700
Re: What load testing solutions do you use? Arne Vajhøj <arne@vajhoej.dk> - 2012-04-03 19:20 -0400
Re: What load testing solutions do you use? Daniel Pitts <newsgroup.nospam@virtualinfinity.net> - 2012-04-03 16:41 -0700
Re: What load testing solutions do you use? Arne Vajhøj <arne@vajhoej.dk> - 2012-04-03 19:44 -0400
Re: What load testing solutions do you use? Daniel Pitts <newsgroup.nospam@virtualinfinity.net> - 2012-04-03 17:02 -0700
Re: What load testing solutions do you use? Wayne <nospam@all.invalid> - 2012-04-05 13:38 -0400
Re: What load testing solutions do you use? Scott Price <scottp@customercentrix.com> - 2012-04-07 10:50 -0700
| From | Daniel Pitts <newsgroup.nospam@virtualinfinity.net> |
|---|---|
| Date | 2012-04-03 11:19 -0700 |
| Subject | What load testing solutions do you use? |
| Message-ID | <%aHer.1879$ie6.974@newsfe14.iad> |
I've been looking for load testing solutions, and there are quite a few choices, many are expensive, some seem less than useful. So I'll ask here. What have you used in the past for load testing? I'm specifically testing Java servlets, which request data from either HTTP XML resources, or databases (primarily MySQL). The difficulty I have is that setting up an entire environment for our system is not a trivial task (large amounts of data, and very complex system). I have three kinds of load testing I'd like to perform. 1. Failure mode testing. What happens when the downstream datasources start to fail under load, what can be done to make that more graceful. 2. Capacity and scalability planing. 3. Performance tuning for individual requests. 2 and 3 we can glean /some/ useful information from our production environment, but we don't want to test failures or cause failures in production, obviously. Thanks, Daniel.
[toc] | [next] | [standalone]
| From | Arne Vajhøj <arne@vajhoej.dk> |
|---|---|
| Date | 2012-04-03 19:20 -0400 |
| Message-ID | <4f7b85c4$0$286$14726298@news.sunsite.dk> |
| In reply to | #13352 |
On 4/3/2012 2:19 PM, Daniel Pitts wrote: > I've been looking for load testing solutions, and there are quite a few > choices, many are expensive, some seem less than useful. So I'll ask here. > > What have you used in the past for load testing? > > I'm specifically testing Java servlets, which request data from either > HTTP XML resources, or databases (primarily MySQL). The difficulty I > have is that setting up an entire environment for our system is not a > trivial task (large amounts of data, and very complex system). I have > three kinds of load testing I'd like to perform. > 1. Failure mode testing. What happens when the downstream datasources > start to fail under load, what can be done to make that more graceful. > 2. Capacity and scalability planing. > 3. Performance tuning for individual requests. > > 2 and 3 we can glean /some/ useful information from our production > environment, but we don't want to test failures or cause failures in > production, obviously. Obviously you need something to call the servlets. There are commercial offerings like LoadRunner, open source Apache JMeter, you can build something yourself on top of Junit or Cactus, you can build something yourself from scratch. To be able to test #1 you need to be able to mock the backend with something you can control responsiveness for. Arne
[toc] | [prev] | [next] | [standalone]
| From | Daniel Pitts <newsgroup.nospam@virtualinfinity.net> |
|---|---|
| Date | 2012-04-03 16:41 -0700 |
| Message-ID | <4VLer.1941$ie6.332@newsfe14.iad> |
| In reply to | #13354 |
On 4/3/12 4:20 PM, Arne Vajhøj wrote: > On 4/3/2012 2:19 PM, Daniel Pitts wrote: >> I've been looking for load testing solutions, and there are quite a few >> choices, many are expensive, some seem less than useful. So I'll ask >> here. >> >> What have you used in the past for load testing? >> >> I'm specifically testing Java servlets, which request data from either >> HTTP XML resources, or databases (primarily MySQL). The difficulty I >> have is that setting up an entire environment for our system is not a >> trivial task (large amounts of data, and very complex system). I have >> three kinds of load testing I'd like to perform. >> 1. Failure mode testing. What happens when the downstream datasources >> start to fail under load, what can be done to make that more graceful. >> 2. Capacity and scalability planing. >> 3. Performance tuning for individual requests. >> >> 2 and 3 we can glean /some/ useful information from our production >> environment, but we don't want to test failures or cause failures in >> production, obviously. > > Obviously you need something to call the servlets. > > There are commercial offerings like LoadRunner, > open source Apache JMeter, you can build something > yourself on top of Junit or Cactus, you can build > something yourself from scratch. There is also apachebench for simple cases. > > To be able to test #1 you need to be able to mock > the backend with something you can control responsiveness > for. Yup, I've seen commercial products that do this (ParaSoft Virtualize is one such thing). Though the license model doesn't really suit small-scale load-testing. We started working on our own home-grown solution, but if there are other solutions to that problem, I'd love to know about them. We're most interested in HTTP backend simulation, we don't need to worry so much about the DB stuff, at least not for my team. Thanks for the reply, Daniel.
[toc] | [prev] | [next] | [standalone]
| From | Arne Vajhøj <arne@vajhoej.dk> |
|---|---|
| Date | 2012-04-03 19:44 -0400 |
| Message-ID | <4f7b8b69$0$286$14726298@news.sunsite.dk> |
| In reply to | #13359 |
On 4/3/2012 7:41 PM, Daniel Pitts wrote: > On 4/3/12 4:20 PM, Arne Vajhøj wrote: >> On 4/3/2012 2:19 PM, Daniel Pitts wrote: >>> I've been looking for load testing solutions, and there are quite a few >>> choices, many are expensive, some seem less than useful. So I'll ask >>> here. >>> >>> What have you used in the past for load testing? >>> >>> I'm specifically testing Java servlets, which request data from either >>> HTTP XML resources, or databases (primarily MySQL). The difficulty I >>> have is that setting up an entire environment for our system is not a >>> trivial task (large amounts of data, and very complex system). I have >>> three kinds of load testing I'd like to perform. >>> 1. Failure mode testing. What happens when the downstream datasources >>> start to fail under load, what can be done to make that more graceful. >>> 2. Capacity and scalability planing. >>> 3. Performance tuning for individual requests. >>> >>> 2 and 3 we can glean /some/ useful information from our production >>> environment, but we don't want to test failures or cause failures in >>> production, obviously. >> >> Obviously you need something to call the servlets. >> >> There are commercial offerings like LoadRunner, >> open source Apache JMeter, you can build something >> yourself on top of Junit or Cactus, you can build >> something yourself from scratch. > There is also apachebench for simple cases. >> >> To be able to test #1 you need to be able to mock >> the backend with something you can control responsiveness >> for. > Yup, I've seen commercial products that do this (ParaSoft Virtualize is > one such thing). Though the license model doesn't really suit > small-scale load-testing. > > We started working on our own home-grown solution, but if there are > other solutions to that problem, I'd love to know about them. We're most > interested in HTTP backend simulation, we don't need to worry so much > about the DB stuff, at least not for my team. I would write the mock stuff myself. I would expect the specific requirements to be so special that I doubt there will be a good COTS solution. Arne
[toc] | [prev] | [next] | [standalone]
| From | Daniel Pitts <newsgroup.nospam@virtualinfinity.net> |
|---|---|
| Date | 2012-04-03 17:02 -0700 |
| Message-ID | <CcMer.19590$Ce4.13827@newsfe21.iad> |
| In reply to | #13361 |
On 4/3/12 4:44 PM, Arne Vajhøj wrote: > On 4/3/2012 7:41 PM, Daniel Pitts wrote: >> On 4/3/12 4:20 PM, Arne Vajhøj wrote: >>> On 4/3/2012 2:19 PM, Daniel Pitts wrote: >>>> I've been looking for load testing solutions, and there are quite a few >>>> choices, many are expensive, some seem less than useful. So I'll ask >>>> here. >>>> >>>> What have you used in the past for load testing? >>>> >>>> I'm specifically testing Java servlets, which request data from either >>>> HTTP XML resources, or databases (primarily MySQL). The difficulty I >>>> have is that setting up an entire environment for our system is not a >>>> trivial task (large amounts of data, and very complex system). I have >>>> three kinds of load testing I'd like to perform. >>>> 1. Failure mode testing. What happens when the downstream datasources >>>> start to fail under load, what can be done to make that more graceful. >>>> 2. Capacity and scalability planing. >>>> 3. Performance tuning for individual requests. >>>> >>>> 2 and 3 we can glean /some/ useful information from our production >>>> environment, but we don't want to test failures or cause failures in >>>> production, obviously. >>> >>> Obviously you need something to call the servlets. >>> >>> There are commercial offerings like LoadRunner, >>> open source Apache JMeter, you can build something >>> yourself on top of Junit or Cactus, you can build >>> something yourself from scratch. >> There is also apachebench for simple cases. >>> >>> To be able to test #1 you need to be able to mock >>> the backend with something you can control responsiveness >>> for. >> Yup, I've seen commercial products that do this (ParaSoft Virtualize is >> one such thing). Though the license model doesn't really suit >> small-scale load-testing. >> >> We started working on our own home-grown solution, but if there are >> other solutions to that problem, I'd love to know about them. We're most >> interested in HTTP backend simulation, we don't need to worry so much >> about the DB stuff, at least not for my team. > > I would write the mock stuff myself. I would expect the specific > requirements to be so special that I doubt there will be a good > COTS solution. Again, ParaSoft Virtualize appears to be very good. It supported many different protocols (HTTP, various DBs, etc...) but just too expensive for what we needed. The type of product exists, but I've only found one example of it, and we don't want to pay that much for it.
[toc] | [prev] | [next] | [standalone]
| From | Wayne <nospam@all.invalid> |
|---|---|
| Date | 2012-04-05 13:38 -0400 |
| Message-ID | <4f7dd89f$0$25261$9a6e19ea@unlimited.newshosting.com> |
| In reply to | #13352 |
On 4/3/2012 2:19 PM, Daniel Pitts wrote: > I've been looking for load testing solutions, and there are quite a few choices, many > are expensive, some seem less than useful. So I'll ask here. > > What have you used in the past for load testing? > > I'm specifically testing Java servlets, which request data from either HTTP XML > resources, or databases (primarily MySQL). The difficulty I have is that setting up > an entire environment for our system is not a trivial task (large amounts of data, > and very complex system). I have three kinds of load testing I'd like to perform. > 1. Failure mode testing. What happens when the downstream datasources start to > fail under load, what can be done to make that more graceful. > 2. Capacity and scalability planing. > 3. Performance tuning for individual requests. > > 2 and 3 we can glean /some/ useful information from our production environment, but > we don't want to test failures or cause failures in production, obviously. > > Thanks, > Daniel. To load and stress test web interfaces, about the easiest tool is Microsoft's free Web Capacity Analysis Tool (WCAT, <https://www.iis.net/community/default.aspx?tabid=34&g=6&i=1466>, formerly called the Web Application Stress Tool). Another good tool is slamd from www.slamd.com, but that isn't under active development anymore. Apache JMeter was recommended by some others too. (Note, I personally don't have experience with these tools.) -- Wayne
[toc] | [prev] | [next] | [standalone]
| From | Scott Price <scottp@customercentrix.com> |
|---|---|
| Date | 2012-04-07 10:50 -0700 |
| Message-ID | <21e0404a-6bc2-4b92-8675-10899316c4cd@t16g2000yqt.googlegroups.com> |
| In reply to | #13352 |
An inexpensive cloud load testing tool is LoadStorm. You don't quantify the load you think you will need, but LoadStorm can deliver up to 100,000 concurrent users or 15,000 requests per second. It isn't a fit for many situations, and there is only one way to know if it will work for your web application - try it. The tool is SaaS, so it is offered as a free account forever with a limit of 25 concurrent users. There is no limit to the number of tests you can run. If it works for your requirements, then you can upgrade with a credit card and pay per test. 1,000 concurrent users is $39.90. http://loadstorm.com
[toc] | [prev] | [standalone]
Back to top | Article view | comp.lang.java.programmer
csiph-web