Compiled vs. Batch Mode Footprint and Capacity

[aaike]

What can you say about the eValid footprint.

We like the idea of running multiple browsers, but how much storage
does each take?

What's the size of an EPI-compiled eValid copy compared with a
copy that runs in batch mode? Is there any storage gain if I compile
my script with the EPI interface and then just run multiple processes?

Aaike M.

[eValid]

What can you say about the eValid footprint.

We like the idea of running multiple browsers, but how much storage
does each take?

What's the size of an EPI-compiled eValid copy compared with a
copy that runs in batch mode? Is there any storage gain if I compile
my script with the EPI interface and then just run multiple processes?

Aaike M.

Thanks for asking. Those are tough questions to answer simply. We'll try to give some guidelines based on our experience. But remember: "Your Mileage Will Vary"...

The main issues you need to address are the size of the eValid footprint, and the machine input/output capacity.

(1) Memory/RAM Requirement

In LoadTest mode eValid playbacks are done with the evalidT (THIN) version, which is as small an eValid browser as we can make it.

The starting RAM is ~20MB per instance. Depending on how many URLs are in the script, and how many times the playback is repeated, and the duty cycle of the script, this footprint can grow to above 50MB.

In Windows all of the processes will eventually run, but you surely want to avoid swapping. Once the machine goes into swapping mode it is pretty uch "all over with".

Machines running 32-bit Windows have 3 GB of RAM address space, so you should plan on using all of that.

We usually plan on 125 eValids per machine, to be safe and conservative. The maximum we've seen on a 3 GB of RAM machine is 230, but at that point the machine is "red lined".

(2) CPU Adjustments

Out of the box Windows machines don't usually run more than 15-20 simultaneously browsers happily.

There's a set of adjustments you need to make:

[url=http://www.e-Valid.com/Products/Documentation.9/Loading/machine.html]
Machine Adjustments[\url]

These manipulate the amount of heap space and the size of the virtual swap spaces.

(3) Input/Output Capacity

Some customers switch to a 1 Gbps LAN rather than the 10/100 Mbps.

But usually it is the I/O capacity of the machine that is the limitation.

In other situations, we have had customers deploy machines with extra high-end I/O cards -- to match 1 Gbps LANs.

(4) Our "nominal machine" is a Cor2 Pentium IV at 2.5 GHz with 3 GBytes of AM, and with the CPU Adjustments done.

Hope this helps you out.

Remember, "Your Mileage Will Vary"...

The eValid Team

[eValid]

What can you say about the eValid footprint.

We like the idea of running multiple browsers, but how much storage
does each take?

What's the size of an EPI-compiled eValid copy compared with a
copy that runs in batch mode? Is there any storage gain if I compile
my script with the EPI interface and then just run multiple processes?

Aaike M.

Further to the above, about whether compiling the tests will improve the situation and we think the answer is no, not really. At least we don't think so...

At the end of the day, the dominant factor is size of the eValid browser image and it's resource requirements. An EPI-compiled example won't change the fact that there's still a full-featured browser in there. That the playback is driven from steps in a C++/CPP program doesn't seem to make that much difference...but may other EPI users can chime in on this one.

== eValid Support