Authenticated user form submission spike testing
- or -
Post a project like this- Posted:
- Proposals: 18
- Remote
- #4181701
- OPPORTUNITY
- Expired
Description
We are part of a live events company, we develop websites that are used in conjunction with live congresses for event information, speaker biographies, and in this example, user voting during speaker presentation. Our use case is that we need to test how well our servers and the platform handle under increased activity of live user interaction
The project we would like help with is to produce a test scenario where 500 users log in to our platform using unique user accounts, navigate to a specific page and all attempt to submit the vote form within a 10 second window.
The test is not to ascertain if users can log-in and vote within 10 seconds, eg, that they vote as quickly as they can after logging in. We are aiming to test that 500 votes can all be submitted within 10 seconds, but the caveat is that they must all be unique authenticated accounts.
Ideally we would like the testing script to be written compliant with K6 testing software, as we have some base knowledge of this and an existing account within Grafana, and once the test is demonstrated, we would need to receive a copy of the script with basic instructions on how we could change the number of attempted votes of the time span, should we need to run a more intense test at a later date
Huw M.
100% (22)New Proposal
Login to your account and send a proposal now to get this project.
Log inClarification Board Ask a Question
-
Before collaborating i want to ask you some questions as i value my clients and give them my best. I will list the questions.
Subject-: some questions to gather more information about the project itself:
1. Can you provide a detailed overview of your platform and its primary functionalities?
2. What specific features or components of your platform do you anticipate will experience the most significant load during live events?
3. How frequently do you typically experience spikes in user activity during live events, and what are the typical volumes of traffic you encounter?
4. Are there any specific performance bottlenecks or pain points that you've identified in previous live events or testing scenarios?
5. What are the key performance indicators (KPIs) or success metrics that you'll be evaluating during the load testing process?
6. Are there any regulatory or compliance requirements that need to be considered during the load testing, such as data privacy or security regulations?
7. Can you provide access to any relevant documentation, such as API specifications or platform architecture diagrams, to help me better understand the technical aspects of the project?
8. Are there any specific scenarios or use cases that you would like to prioritize during the load testing, such as user registration, content delivery, or transaction processing?
9. Do you have any specific expectations or requirements regarding the scalability and resilience of your platform under load?
10. Are there any particular testing environments or setups that you recommend for accurately simulating real-world conditions during the load testing?
Huw M.21 Mar 2024- Our platform is designed to be an accompanying website for in-person live events, they are designed to provide information associated with the events, such as biographies, agenda information, livestreaming to in-room participants
- For most of the live events, the users will not be interacting on the platform directly, aside from short windows where they will be asked to vote on specific poll questions, which is what this test is aiming to investigate
- The sharpest spikes are during initial login and during the live presentations where voting is being recorded, our average pax is roughly 500, but there have been instances of 10k upwards, we are aiming to understand a baseline figure for how well our infrastructure can handle requests
- As mentioned, the most pain points come during login/registration and voting
- Not for the purposes of testing, all the testing will be done against dummy content
- We have 2 platforms that will need to be tested upon, they will function the same, however, our primary live environment charges for activity, so during the build of the testing script, we will be asking to do the development on a duplicate platform, before switching to the real environment once it is configured
- Not as much, we are primarily interested in knowing the limitations of the server during the voting
- No, if the infrastructure cannot handle the load, then that will be the outcome of the testing, even though it 'fails' that would still be considered as completing the work
- Not sure I understand the questionLusan S.21 Mar 2024Thank you for providing detailed information about your platform and testing requirements. Based on your specifications, I will create a robust test scenario compliant with K6 testing software. My goal is to simulate 500 unique authenticated users submitting votes within a 10-second window. I'll ensure the script is adaptable for future adjustments in test intensity. Let's proceed with developing the testing script on the duplicate platform as discussed. Looking forward to assisting you further.
-
Is the user voting system functionally complete - unfixed issues in the system could throw the performance testing results off & some functional bugs only manifest at high user loads.
1. Where is the system to be tested hosted - in the cloud or in-prem - are there any limitations with respect to hardware capacity or network traffic ?
2. Has the performance testing environment been sized for 500 users ?
3. My understanding is that the performance test is for 500 authenticated virtual users voting concurrently within 10 seconds - are the virtual accounts for these 500 users available ...
4. ... or, do these 500 virtual users need to be generated for the performance test - if yes, this may need writing a script for as well.
5. What is the authentication mechanism for the 500 virtual users - OAuth2 ?
6. Does the client-side of the user-voting system (browser side) involve any JavaScript or widgets for non-rendering (not UI related) tasks such as encryption, hashing, etc. ?
6. What metrics (client side as well as server- side) are expected out of this performance test ?
7. Is there a "go-live" date involved, i.e. is there a time constraint for the performance test ?
Huw M.21 Mar 2024- There are 2 duplicate systems to be part of the testing, the 'live' hosting provider we use is Acquia for Drupal hosting, except we are charged by requests to the server, so we have a duplicate of the site hosted on a temporary platform where the script can be developed and tested so that we are not running up costs.
- No, the environment is at it's default capacity, we are trying to ascertain whether or not special measures need to be taken into account if the testing demonstrates that 500 is too many within the spike test
- All 500 accounts can be created ahead of time for you, we can import a list of dummy users and provide you with the login information
- The usual login method generates a cookie on the users device related to their session information, the normal login method is Drupal's standard username/password auth
- From our initial test we believe that the poll is generated on the DOM when the page loads, as the HTML does not appear to be there instantly when the page loads. The normal voting mechanism uses AJAX submission
- The results we are aiming to achieve will be to understand whether or not our infrastructure is capable of recording all 500 votes, and whether or not there are server-side issues related to that kind of bottleneck. We are not expecting any failures of this to be resolved as part of the test, we are simply aiming to understand the limitations
- Not an upcoming deadline, this is for internal checking, so timelines are a lot looser -
Could you provide more details on the server architecture and any existing load balancing solutions? Additionally, are there any specific metrics or thresholds you're targeting for server performance under load? This information would be crucial for tailoring the spike test to your system's specifics.