Forum Discussion

Freebooter_1399's avatar
Freebooter_1399
Icon for Nimbostratus rankNimbostratus
Oct 01, 2014

BIG-IP management of client requests queue.

Greetings,

 

I am new to that product and having some troubles in understanding a few things. Does BIG-IP has queue for client requests (especially through api), and if a large number of requests simultaneously can lead to BIG-IP crash or loss of some of these requests?

 

Meaning, my application can make 500+ requests to F5 machine at the same time, using iControl api. Of course in that moment F5 machine will be already heavily loaded with requests from other applications and users. What could be an outcome here, except waiting some time for responses from F5?

 

Will very appreciate any help on a topic. My first post on DevCentral, so constructive critic is welcome. Thanks!

 

application delivery
design
devops
iControl
LTM
management
performance

4 Replies

Sort By
  • Jason_40733's avatar
    Jason_40733
    Icon for Cirrocumulus rankCirrocumulus
    Oct 01, 2014

    The BigIP is a name for a set of hardware (and software now that a virtual version is available). It has many modules that can be licensed per BigIP. These modules do vastly different things.

     

    The BigIP iControl API is for managing, changing, querying the BigIP system. It is not a module itself. Here is a list of the modules that perform functions for environment. https://f5.com/products/modules

     

    Here is a link to an iControl API white paper. It is older but has relevant information letting you get an idea of what iControl is used for. http://www.f5.com/pdf/white-papers/icontrol-wp.pdf

     

    Jason

     

  • Freebooter_1399's avatar
    Freebooter_1399
    Icon for Nimbostratus rankNimbostratus
    Oct 01, 2014

    I am using Local Traffic Manager module, and already successfully using iControl in my .net based application. My concern is performance, what will happen if large numbers of requests sent to LMT simultaneously from my app. Want to avoid LMT crash or lost requests, and at the moment can't simply test that case.

     

  • What_Lies_Bene1's avatar
    What_Lies_Bene1
    Icon for Cirrostratus rankCirrostratus
    Oct 02, 2014

    There is a session/connection limit for iControl as well as idle timeouts etc. These seem to be hard to control but regardless, I can't conceive of anyone managing to crash LTM via iControl load. iControl processing takes place in the HMS (the management subsystem) and LTM runs within TMM. The two are coupled but at worst you would slow down the management system, not LTM functions.

     

    I'm sure sending 500 requests may result in delayed responses (or drops even) but only in relation to iControl. Code appropriately.