posted 02-07-2003 03:08 AM         

I would like to know the following:

1) What is performance testing

2) In India , are the companies doing performance testing?

3) What is the min. requirement to do performance testing?

4) What is the min. qualification required to do performance testing?

5) Suggest some books to improve the knowledge about performance testing?

Regards,
R.Deiva

posted 02-07-2003 05:43 AM         

And I would think a simple internet search would reveal the answer to 2.

------------------
Scott Barber, Sr. Performance Engineer
sbarber@noblestar.com http://www.noblestar.com http://www.perftestplus.com

posted 02-09-2003 05:12 AM         

quote:

---

Originally posted by RSBarber:
And I would think a simple internet search would reveal the answer to 3.

---

posted 02-10-2003 02:56 AM         

quote:

---

Originally posted by Deivasigamani:

2) In India , are the companies doing performance testing?

---

Everyone in the big-8 consulting category have a performance engineering service now.
This includes:
InfoSys (Service called Infolater ??), TCS(located in Mumbai I think - called Enterprise Performance something), Cognizant(Chennai & Pune), HCL, Polaris(Chennai).
Most mid-size players like Kanbay(where I'm at) are starting as well - believe we are behind the curve by a year and a half.

quote:

---

3) What is the min. requirement to do performance testing?
4) What is the min. qualification required to do performance testing?

---

To be a Performance Analyst you must :
*Know queuing theory well.
*Have a good understanding of mathematics.
*Be conversant with statistics too.
*Programming skills needed. A background in software testing prepares you mentally.
*Be good at deriving useful information from a deluge of test results.

An engineering background is what I highly recommend.

posted 02-10-2003 04:51 AM         

While I agree with you about the importance of those things (and I am going to add them to the existing list when I find it ), I don't know if that is the level of detail the poster is talking about. You really don't need any of those things to conduct a performance test - you need some of them to analyze performance test results, and all of them (and more) to tune the system based on the results. (ok, I'll hedge my bet a little - one needs to at least understand the basics of queueing theory to develop an accurate test)

Again, not challenging your answer, just not positive that was what the poster intended in their question.

posted 02-10-2003 06:21 AM         

Tasks are:

1. Usage profiling/requirements gathering.
* Good analysis and data collection skills.

2. Test design.
* Knowledge of technology involved - web servers, app server, DB servers etc.
* Queuing theory definitely helps.

3. Test scripting and debugging.
* C/Pascal/VB - programming exp helps - depends on what tool you use.

4. Test execution.
* Ok, here's where minimal skills are needed.

5. Result analysis.
* Statistics - atleast you should be able to understand what resources are characterized by what statistical entities.
*Mathematics/Calculus helps understand teh relationship between various variables.

6. Bottleneck detection & tuning
* Advanced architectural knowledge.
* Analysis skills
* Knowledge of OS/DB/networks

Which is why I recommended an engineering background.

I find that most folks in performance engineering are not really capable of many of these tasks. The biggest fallacy propagated is the equating of tool scripting knowledge to the entire thing. You should subscribe to the LoadRunenr mailing list. So many folks know the tool but are clueless when it comes to using the results intelligently.

posted 02-10-2003 08:59 AM         

Symantics, and I'm being picky, but most of the "world" seems stuck on testing, and isn't interested in the rest... "my developers can find and fix the problems, just give me numbers - I'm not paying you to do what my developers should already be doing." I can't tell you how often I hear that... or...
"I can get a free load test from Mercury - I don't understand why what you do is worth more than free."
Drives me insane, but I'll get off the soapbox and just add this link to the faq as well.

posted 04-04-2003 06:28 PM         

I don't know much (relatively speaking) about queueing theory - I've only had 1.5 graduate level courses on the topic.

Basically it's the mathmatics behind how distinct items move through a choke point (simplification) - For instance, it's the mathematical model that can predict how traffic will navigate through a multi-lane toll booth during rush hour. The same mathematical models apply to requests, packets, threads, etc.

I don't know where to go for "quick and easy" information. Sorry.

posted 04-09-2003 04:16 AM         

quote:

---

Originally posted by Deivasigamani:
Everyone in the big-8 consulting category have a performance engineering service now.
This includes:
InfoSys (Service called Infolater ??), TCS(located in Mumbai I think - called Enterprise Performance something), Cognizant(Chennai & Pune), HCL, Polaris(Chennai).
Most mid-size players like Kanbay(where I'm at) are starting as well - believe we are behind the curve by a year and a half.

---

I am told TCS has relation with Mercury. Also all the US companies with bases in B'lore are likely to have it.

Just wondering what kind of things they perf test. Do they take specific perf testing projects or is it the part of testing process for their own projects? Don't the US clients like to see it being done in front of them rather than being done from India?

Punekar one thing I would like to ask you. You said that you believ that you are one and a half year behind the curve. What kind of difference could that cause? Would it be in terms of skills of employees or the process maturity or the infrastructure facilities?

One thing about Performance testing skills. I believe perf testing is not for QA dept. The problem is however that it has the word 'testing' in it. You cannot make test scripts,matrix, documents, automation scripts, find 100% of the bugs in functional testing and yet find time to become a well rounded performance test professional. Where will they find the time to know about session management, cookies, javascripting, c, sytem counter interpretation, database system (V$) table value interpretation, load balancing, clustering, SSL, firewall, NTLM authentication, data compression, garbage collection anlaysis, capacity planning etc?

An unfortunate problem with perf testing is that almost any script will run and produce some numbers and that will convince the QA Manager that the job is done. What happens is that as that as it mostly takes time for the load to build and by the time the project people are gone and thus there is no accountability.

regards,
Cvik

posted 04-09-2003 04:50 AM         

quote:

---

Originally posted by cvik:

One thing about Performance testing skills. I believe perf testing is not for QA dept. The problem is however that it has the word 'testing' in it. You cannot make test scripts,matrix, documents, automation scripts, find 100% of the bugs in functional testing and yet find time to become a well rounded performance test professional. Where will they find the time to know about session management, cookies, javascripting, c, sytem counter interpretation, database system (V$) table value interpretation, load balancing, clustering, SSL, firewall, NTLM authentication, data compression, garbage collection anlaysis, capacity planning etc?

An unfortunate problem with perf testing is that almost any script will run and produce some numbers and that will convince the QA Manager that the job is done. What happens is that as that as it mostly takes time for the load to build and by the time the project people are gone and thus there is no accountability.

---

I've been making this arguement for years. Along with the one that Configuration Management isn't a QA Role. I usually lose. Within my own company I have finally gotten the project team org chart to show the Sr. Performance Engineer as a peer to the QA Manager AND the Sr. Architect (technical lead). This person is responsible to the project manager. The reason is that the dev team and the QA team generally have very different needs/expectations of the performance team - this way the Sr. Performance Engineer and the Project Manager can prioritize those needs/expectations. When performance falls under QA, it gets focused on numbers and not helping the developers tune their stuff. When it falls under development, the requirements verification part gets lost in the middle of all the developers trying to perfect their part of the code.

NE-way - long winded way of saying "I agree".