Nobody likes paying for something that could have been avoided. Here are some pro tips for making sure you get the most out of your Flood subscription:
1. Start small.
One mistake we often see is a new flooder running 40 nodes with 1000 users each as a first test... only to realise that the first request isn't right and is failing 100% of the time. As much as we appreciate your patronage, do yourself a favour and take baby steps.
Run with a single user on a local instance and debug with it until all the scripting errors have been sorted out. Then run several iterations with two users. If that still looks clean, do a shakeout with a few users on Flood with one node.
Doing it this way will give you plenty of opportunity to debug issues before you start paying for node hours.
2. Utilise the node's full resources.
Once you've got all the errors ironed out and you've done a few shakeout test at increasingly higher levels, it's time to start running tests. One thing to consider is that we don't arbitrarily limit the number of virtual users you can run on a node, but there are still restrictions based on what the node is capable of.
As a general rule, we recommend that you run about 1000 users if you're using JMeter/Gatling (check out our article on calculating how many nodes you'll need). To eke out as much value as possible from each node, try to run as many users on each node (up to its limit) before starting new ones, if possible. However, it's always a good idea to keep an eye on the resource utilisation of your node to see if it's getting overtaxed.
You don't want to run so many users on a node that it starts to struggle, introducing delays into your test that have nothing to do with your application under test.
3. When launching a grid, set an appropriate duration for it.
If you want to run a five-minute test and don't anticipate running any other tests, don't set the grid to run for an hour.
Flood charges you based on how long you have grid nodes up, and not how long your test runs.
4. Stop flood/grids when you're not using them.
If you start a flood or a grid and change your mind, you can always stop them after the fact. It's good practice to do a quick check to see if there are any running grids before you log out for the day, just in case there's a grid you've forgotten about.
5. Maximize the 12-minute window.
The minimum amount of time that you'll be charged for is 12 minutes. This means that if you run a test for five minutes and run nothing else, you will still be charged for 12 minutes. So why not make the most of it? You can get three four-minute tests in a 12-minute period. You can also use this trick to plan your testing in multiples of 12 -- instead of running a 20-minute test, consider running two 12-minute tests for the same cost.
6. Keep an eye on your node hour balance.
This might seem obvious, but it's often overlooked - you can see your node hour balance on the toolbar:
If it says "X hours left", that means you haven't yet reached your subscription quota for the month.
If you click on your balance, you'll see your grid usage history, which can help you to reconcile your usage with your bill.
7. Cancel your subscription if you're not going to be testing for a while.
We're always sad to see you go, but we totally understand that testing needs sometimes flow and ebb during the year. If you're not going to need Flood, don't forget to cancel your subscription before you're billed for the next month. Just note that at the end of your billing period, you will lose all remaining node hours, so be sure to use them before that occurs. You can check the date that your billing cycle ends here:
If you ever have any questions about how our pricing system works and how to get the most of your subscription, please let us know. Send us a message through our live chat or through email@example.com and we'll be glad to help.