Question about BlazeDS Channel internals

There is much useful information available about how an application developer would use BlazeDS to implement their application, but there is a shortage of information about how BlazeDS is implemented.  There are very general architecture diagrams, but no information on the actual network technologies employed.  This information is essential for forming estimates of the server hardware requirements of a design.  The alternative is to build the application one way, and then discover the performance characteristics afterwards.
For one feature of my application, the ideal structure would be to give the clients and mx.Consumer to receive non-polling stream of changes replicated from the database.  In this particular feature, there would be no subtopics, each client would receive the same data.  The question is would BlazeDS/Flex give each client 1 persistent socket to support the stream.  If this is the case, then this design poses a very significant load on the server.  It makes it easy for the developer, because he writes to only one topic, but if under the covers, the server is writing to one socket for each client.  It will require a lot of hardware.
Or, to state the question another way, would BlazeDS/flex use multicast, when there is a client Consumer subscribing to a topic, with no subtopics?

I have just found one clear statement in the BlazeDS documentation with regard to sockets when streaming:
"The Servlet API uses blocking IO so as with long polling above, you must set a configured upper bound on the number of streaming connections you allow. Clients that exceed this limit are not able to open a streaming connection and will fall back to the next channel in their channel set."
BlazeDS Developer Guide, page 50.
This I did not expect, since non-blocking IO has been part of Java for over 5 years in the NIO package.  If I were to deploy BlaseDS on an Intel sized box, I suspect the blocking IO would limit me to maybe 300 concurrent users.  If this is true, it means that both streaming and long polling are unusable for almost any application on the public web.
Maybe the only option is to write a server in Java NIO that communicates with Flash Sockets.
Did I miss something?

Similar Messages

  • Question about blazeds turnkey, tomcat and directory structure

    hi. this question is pretty basic...been reading sujit reddy g's blog on installing/setting up blazeds.
    in one article he creates a samplewebapps directory in C:\Program Files\Apache Software Foundation\Tomcat 6.0\webapps\samplewebapps and copies the blazeds WEB-INF/lib into that directory and the configuration files in the flex folder across as well...http://sujitreddyg.wordpress.com/2009/04/07/setting-up-blazeds/
    in another article on invoking java methods from flex he configures the remote-config.xml file directly in the blazeds\WEB-INF\flex folder....http://sujitreddyg.wordpress.com/2008/01/14/invoking-java-methods-from-adobe-fle x/
    wasn't sure why in the first example he copied the files and folders to the samplewebapps directory while in the second example he just configured the files within the blazeds directory...thanx...(i'm a newbie at server side development)

    I'll take a stab at it. The key thing to realize is the BlazeDS code is ADDED on
    to the appserver. E.g. for Tomcat/WebLogic/et al one adds the reference in the web.xml file in WEB-INF.
    So, what is that add-on?
    1. Executable files. These are jar files and typically stuck into WEB-INF/lib
    2. Configuration files. flex/services-config.xml is specified in web.xml. It refers to the other config files in WEB-INF/flex
    So, the config in web.xml tells Tomcat (and its forked commercial products) to load up the Flex jars and run some classes. By standard, the "run some classes" follows the servlet lifecycle and runs specific methods in the class when the servlet is loaded, called, destroyed. So, Flex jars have a class which implements the servlet interface.
    Incidentally, you may also see references to log4j, Spring, and other frameworks in the web.xml as well. They do the same sort of stuff. So, Tomcat does the passing of the HTTP packets and stages them into Java classes and the hooked in frameworks do add their own behaviours to the setup.
    HTH,
    TimJowers
    P.S> Also note in Flex when you setup the project properties for a Flex Project then you need to match up your URL and "context" to what you have on your server. In his exampe, the "samples" context may have already been setup so easier to use. What is a "context"? The idea is to have more than one webapp running on an appserver. In Tomcat, its basically just the subdirectory under "webapps". That directory name becomes part of the URL. E.g. webapps/samples -> http://localhost:8080/samples  or webapps/mytest -> http://localhost:8080/mytest

  • Question about blazeds turnkey installation and directory structure

    hi. this question is pretty basic...been reading sujit reddy g's blog on installing/setting up blazeds.
    in one article he creates a samplewebapps directory in C:\Program Files\Apache Software Foundation\Tomcat 6.0\webapps\samplewebapps and copies the blazeds WEB-INF/lib into that directory and the configuration files in the flex folder across as well...http://sujitreddyg.wordpress.com/2009/04/07/setting-up-blazeds/
    in another article on invoking java methods from flex he configures the remote-config.xml file directly in the blazeds\WEB-INF\flex folder....http://sujitreddyg.wordpress.com/2008/01/14/invoking-java-methods-from-adobe-flex/
    wasn't sure why in the first example he copied the files and folders to the samplewebapps directory while in the second example he just configured the files within the blazeds directory...thanx...(i'm a newbie at server side development)

    There's really not much difference. In some cases, you might already have a web application you are using so you could just copy the contents of the blazeds web app into your existing web application.
    Application servers can host more than one web application and a Flex application that uses BlazeDS can be deployed on a different web app than the one where BlazeDS is running or a different server entirely so it really just depends how you want to set things up.
    Hope I'm not just making things more confusing for you. . .
    -Alex

  • This is a very simple question about multi-channel audio playback

    I have an mp4 file that i made and i made it 7.1 surround sound, and i'm pretty sure that this 7.1 surround sound works, as it can be played in VLC. i'm using netstream to load my files now. How do i make it so that flash can playback all 8-channels of sound? I suppose kglad would know the answer

    Flash doesn't support multi-channel audio (yet?)

  • Question about BlazeDS on a remote machine

    So, i am running BlazeDS on my windows machine because that is where i have my Java stuff all setup and entrenched.
    However, i develop Flex on my mac because i like how it works, and i have Flash Builder on Mac.
    I want to set up a BlazeDS project, but it's asking for the root installation folder. How can i do this if it is on a remote machine?
    The first answer that comes to mind is to map the Tomcat folder on my mac, but i don't know if that will work right.
    any ideas/suggestions?
    thanks

    bumpity bump....

  • Bursting - question about delivery channel

    Can I call scripts with one of the delivery channels? Or maybe I can do it through filter, something like
    <xapi:filesystem output = "/tmp/1.pdf" filter="/tmp/myscript.sh /tmp/1.pdf">
    ??

    Many reasons. For exampel one of our request is to call web-services for some bursted file. If I could call a script this will become doable.
    You mentioned BurstingStatusListener . I didnt' find any reference to BurstingStatusListener in Administration guide. Can you give some example?

  • Question about Logic Channel EQ

    When i use Logic Channel EQ to cut low frequencies of say, a kickdrum (in this particular case up to 200hz), the level meter on the channel increases quite a bit. I thought cutting frequencies is supposed to reduce the volume? Also there arent any other plug ins on the channel. Is this normal?

    At it's default setting the low cut on the channel eq has quite a bit of filter resonance at the cutoff point. This resonant boost is what you are hearing. If you lower the Q all the way this resonance will be gone and you will get only the cut you are looking for. If you use the graphic analyzer be sure to set it to maximum resolution as it is quite misleading otherwise.

  • Question about audio channels in FCP

    Because of the way audio was captured with a few clips in my sequence, the clips audio is only coming through one channel (speaker). The other channel contains nothing. How can I duplicate the channel with audio and assign it to the other speaker?

    I did notice that. There are a small few of us on here.
    So now that I've helped you out, bring me some Great Plains Sauce & Dough.

  • Questions about 3702i and WSM (WSSI)

    I have several questions about the 3702i AP and the WSM that do not seem to be answered in the data sheets/documentation.
    Any insight would be appreciated.
    1. Is the WSM officially supported on the 3702 AP? The documentation seems to conflict on this.
    For example, at the top of the WSM data sheet: "Flexible add-on third radio module for the Cisco Aironet 3600i or Cisco Aironet 3600e Series Access Points". Then later on in the page graphic shows the 3700 series radios too. I have found multiple pages that have similar conflicting information.
    2. Can the WSM be used WITHOUT MSE and Cisco Prime? If so, are there any limitations?
    3. How does the licensing work? Where is the license installed -- WLC, MSE, Prime? Do ANY of the features work without the license?
    3a. How do you order the license? The SKU listed on the data sheet (L-WIPS-MM-1AP) is not a valid part number in CCW.
    4. Is it possible to do sniffing with a WSM? (the idea being not having to dedicate an AP to sniffing, or taking down a client-serving AP to sniff)
    Thank you.

    Ok, so you are talking about off-channel scanning, which is actually a very short duration 50ms, that the AP is actually off-channel to scan.
    http://mrncciew.com/2013/03/16/configuring-rrm/
    Are you thinking of this bullet point?
     Access points with wireless security module: The Cisco Aironet® Access Point Module for Wireless Security aggressively scans all channels without affecting the data-serving radios in the 2.4- and 5-GHz bands.
    To me, that's just saying that the WSM is scanning and not interferring with the AP's normal radios.
    HTH,
    Steve

  • Some questions about a Time Capsule

    Hi!!
    I just have a few questions about a Time Capsule, but first let me explain the backgroud to my questions.
    I live in a 2 floor house(including a basement) and my bedroom is downstairs. My parents currently have a desktop PC connected to our wireless router and our printer. Me and my step-father both have 13 inch MacBook Pros, and he uses the Mac like a basic computer, so the equipment connected to our PC is good for him. I, on the other hand, use my Mac, iPad and iPhone to their fullest potential. I am a SUPER MEGA NERD when it comes to computers, so I plan on making my bedroom an amazing office at the same time. So I want an AirPrint Enabled printer in my bedroom (Under 100$$) along with a Time Capsule to back up my information, because everyday I look around in the backdoor of my Mac alot, and if I jack anything up (And I cant handle the problem) I would like to restore my Mac to a previous version. So here are my questions:
    1) My bedroom is right underneath our router, so i get GREAT Wi-Fi in my room. If I get a Time Capsule, can I use it in my room to extend my Wi-Fi, and how would that work? Will it extend my wifi under the same name, because I would like my own Wi-Fi (Titled to what I want with my own password)
    2) If I have to create a NEW Wi-Fi network, how would I do that?
    3) Would I need to connect my new printer directly to my ROUTER, or could I connect it to the TIME CAPSULE?
    If you don't understand what I'm trying to ask (Even though I feel like it's self-explanatory)
    ONE question about the PRINTER: Do any of you have any recommendations for an AirPrint Enabled printer under 100$? Even under-150$ would be okay.
    THANKS FOR ANY HELP!!!!

    everyday I look around in the backdoor of my Mac alot, and if I jack anything up (And I cant handle the problem) I would like to restore my Mac to a previous version. So here are my questions:
    Don't just depend on TM.. use a disk image type utility.. either one of the Mac tools, SuperDuper, CCC, Chronosync or even a linux boot cd (if you have a cd/dvd drive still or via usb can boot) a great tool is Clonezilla.. it will do basically any computer, Mac, Linux, Windows.. and since it boots into its own OS can restore a computer in a flash. Then you can add the incremental changes from TM.
    1) My bedroom is right underneath our router, so i get GREAT Wi-Fi in my room. If I get a Time Capsule, can I use it in my room to extend my Wi-Fi, and how would that work? Will it extend my wifi under the same name, because I would like my own Wi-Fi (Titled to what I want with my own password)
    Actually I am surprised you get good wireless as usually directly above or below a router is often the worst spot. Time Capsule cannot extend the wireless of any other router than Apple.. it is a special system and only Apple can join the club. It has bad effect though, if you set the TC to extend wireless it turns off ethernet ports.. which is hugely dumb.
    The best arrangement is ethernet to the upstairs.. or EOP adapters. If not then a wireless router that matches the existing unit will generally be better than mixing apple and non-apple.
    2) If I have to create a NEW Wi-Fi network, how would I do that?
    3) Would I need to connect my new printer directly to my ROUTER, or could I connect it to the TIME CAPSULE?
    You create a new wireless network by selecting manually the wireless channel and use different wireless name, ssid and security. There are 3 available channels on 2.4ghz.. your existing router will use one or two.. depending on if the setting is 20mhz or 40mhz.. so you need to manually set everything and not use auto.
    TC is slow on 2.4ghz cf 5ghz and there are many more channels at 5ghz but not everything can use it.. iphone for instance is 2.4ghz only . whereas ipad can use 5ghz.. most full computers now can use 5ghz.
    Buy a wireless network printer. Or alternatively an ethernet network would be also ok but now are more expensive than wireless ones. Network printers are not much more than straight USB and work much better.
    I do not use airprint so have nothing to offer.. find the best HP model in your budget.. but others may have had more experience.. I use ethernet connected b/w laser business type HP.. they work great.. but airprint is too recent an innovation for me.

  • Two Questions about Airport Express

    All, I am a recent convert to using a Mac so bear with me.
    I have two questions about the Airport Express.
    Question if I buy one in the USA, I will visiting this week on holiday,can I use it the UK with a UK adaptor/apple plug?
    Can the Express be used with my current wireless network, using a Netgear DG834G V4, I have read lots in the forums about putting a $ sign in front of the WPA password?
    Any help and advice gratefully received.

    Tesserax wrote:
    Hello andixbox. Welcome to the Apple Discussions!
    Question if I buy one in the USA, I will visiting this week on holiday,can I use it the UK with a UK adaptor/apple plug?
    Yes, but at least two things to consider:
    o The US version of the 802.11n AirPort Express Base Station (AXn) only supports radio channels 1-11 in the 802.11g radio mode. Channels 12-14, used in other countries outside the US will not be available.
    o The AXn supports both the 2.4 and 5 GHz radio bands for 802.11n. The 5 GHz band is restricted in certain countries (like the UK) so operating in this mode may get you unwanted attention.
    That should be okay then because the Netgear broadcasts on channels 1-11. I hadn't appreciated the 5 Ghz band, but having checked both US and UK Apple sites the techspecs both mention 2.4 Ghz and 5 Ghz.
    Can the Express be used with my current wireless network, using a Netgear DG834G V4, I have read lots in the forums about putting a $ sign in front of the WPA password?
    If your intent is to have the AXn join the Netgear as a wireless client, then yes it should work. The "$" requirement was for WEP. This shouldn't be necessary for WPA or WPA2.
    I intended to use the AXn as an range extender, but you mention wireless client, is that something different?

  • Question about the programming of a legend

    Hello everybody,
    I have a question about the programming of a waveform's legend. I
    already asked here in this forum about the legend programming (03)
    months ago.
    I went satisfied but I ve just noticed that this code
    (See Code old_legend_test.llb with main.vi as main function) operates a
    little different from my expectances.
    Therefore I have a new question and I want to know if it
    is possible by labview programming to plot and show, on a waveform
    chart, a signal with activ plot superior to zero (0) without to be
    obliged to plot and show a signal with activ plot equal to zero (0) or
    inferior to the desired activ plot.
    I give you an example
    of what I m meaning. I have by example 4 signals (Signal 0, 1, 2 and 3)
    and each signal corresponds respectively to a channel (Chan1, Chan2,
    Chan3, Chan4). I want to control the legend (activ plot, plot name and
    plot color) programmatically. Is it possible with labview to plot signal
    1 or 2 or 3 or (1, 3) or (2,3) or (1,2,3) or other possible combination
    without to active the signal with the corresponding activ plot zero
    (0)?
    Let see the labview attached data
    (new_legend_test.llb with main.vi as main function). When I try to
    control the input selected values again I get them back but I don't
    understand why they have no effect on the legend of my waveform chart.
    Could somebody explain me what I m doing wrong or show me how to get a
    correct legend with desired plots? Thank by advance for your assistance.
    N.B.
    The
    both attached data are saved with labview 2009.
    Sincerly,PrinceJack
    Attachments:
    old_legend_test.llb ‏65 KB
    new_legend_test.llb ‏65 KB

    Hi
    princejack,
    Thanks for
    posting on National Instruments forum.
    The behavior
    you have is completely normal. You can control the number of row displayed in
    the legend and this rows are linked to the data you send to your graph. Thus,
    if you have 3 arrays of data, let say chan1, chan2 and chan3, you can choose
    which data you want to display in your graph using the property node (Active
    plot and visible). But for the legend as you send 3 plots there is an array of
    the plot name [chan1, chan2, chan3] and you can display 0, 1, 2 or 3 rows of
    this array but you cannot control the order in this array! So, to be able to
    change this array you have to only send data you need to you graph. I'm not
    sure my explanations are clear so I have implemented a simple example doing
    that.
    Benjamin R.
    R&D Software Development Manager
    http://www.fluigent.com/
    Attachments:
    GraphLegend.vi ‏85 KB

  • Question about swapping SIM cards on a Z10

    Hi, everyone, I am a non-Blackberry user who works in IT support.  One of my clients uses a Blackberry and has often asked me very complicated questions about how to do various unusual things with his device.  At the present time, he is considering switching to a Z10, which would be set up using ActiveSync.  He wants to know if he could swap the SIM card without having to go through the setup process again.   My client travels a lot, and wants to use a local SIM card on an upcoming trip to a very remote area.  
    I've not been able to find an answer to this anywhere. 
    This is his exact question: "I just got it from a friend (ATT version) and I am trying to see if I like it enough to switch to it as my primary phone.  Blackberry z10 asks you to sign in with your blackberry world ID when you first set up the phone.  Question: Would you have to do it again if you put a different simm into it?  With iPhone, it makes no difference so long as the device is unlocked—you can then exchange simm cards as often as you want.  Would it be the same with an unlocked blackberry z10 (I know it's not possible with the bold 9900 as System 7 ties your ID to the simm card).  In short, in blackberry z10, could you exchange the US simm for another country's simm without having to wipe the phone,  etc.  "
    If anyone can shed some light on this for me, I'd be very grateful.,  Thanks!

    Hi and Welcome to the Community!
    The key issues are thus:
    Ensuring that the device is truly carrier unlocked
    Ensuring that the specific device model is compatible with the radio frequencies/technologies that the desired carrier uses. There are 4 different Z10 models:
    http://www.bbin.in/en/2013/03/03/4-types-blackberry-z10-models-fits-better/?utm_campaign=4-types-of-...
    If both of those are OK (separate research of the desired carriers would of course be necessary), then it is indeed a simple matter to change carriers. For example, I changed from ATT to TMO by simply changing the SIM...no need to WIPE or anything, and my same BBID came along just fine. You should, however, always obtain a properly-sized SIM card from the carrier -- there are plenty of reports of folks trying to use "cut" SIM cards and experiencing errors.
    With BB10, it's actually simpler than with legacy BBs. Legacy BBs used the BIS service, which required receipt onto the device (from the carrier data network) of Service Books so as to activate the various services that the user subscribed to. When changing carriers, it often could be that a latent Service Book from the prior carrier causes issues when shifting to a different carrier...and that's where WIPING and such came into play, so as to cleanse the device so that the new carrier could deliver all of the proper Service Books to the device. With BB10, there is no BIS, and there are no Service Books...so all of that complication is eliminated.
    But, as always, YMMV.
    Good luck!
    Occam's Razor nearly always applies when troubleshooting technology issues!
    If anyone has been helpful to you, please show your appreciation by clicking the button inside of their post. Please click here and read, along with the threads to which it links, for helpful information to guide you as you proceed. I always recommend that you treat your BlackBerry like any other computing device, including using a regular backup schedule...click here for an article with instructions.
    Join our BBM Channels
    BSCF General Channel
    PIN: C0001B7B4   Display/Scan Bar Code
    Knowledge Base Updates
    PIN: C0005A9AA   Display/Scan Bar Code

  • Question about setup_MITE in the PCI-MIO-16E-1

    I have a quick question about setup_MITE in the PCI-MIO-16E-1. In one of the manuals, it states that if you're using Dell Optiplex GX series for our hardware configuration, is it true we need to re-write the MITE functions to correlate with the hardware? Please if you can clarify what this means, I would really appreciate it. Thank you in advance.

    National Instruments is starting a new newsgroup to provide a forum for customers to discuss register level programming issues and ask questions. This newsgroup is not a direct support channel for register level programming, but may be monitored by National Instruments engineers. Please post you question with additional detail in:
    natinst.public.daq.driver-development.general
    Christian
    Christian Loew, CLA
    Principal Systems Engineer, National Instruments
    Please tip your answer providers with kudos.
    Any attached Code is provided As Is. It has not been tested or validated as a product, for use in a deployed application or system,
    or for use in hazardous environments. You assume all risks for use of the Code and use of the Code is subject
    to the Sample Code License Terms which can be found at: http://ni.com/samplecodelicense

  • Question about the "iPod radio remote"

    I assume the "radio remote" is the same for every country and is under software control. I want to buy in Canada but use in UK and Canada.
    Has anyone got one and seen the RDS work, in the UK almost every FM channel is RDS. Do they have RDS in Canada (Toronto) can anyone confirm?
    And of course it says FM so this again I assume no AM, not that I need AM it was just an interest in if it could do it.
    What happens about headphones, you can plug a pair into the remote, can you still plug into the bottom of the iPod or does the remote connector get in the way.
    What is the story with "line out" I use this all the time as I amp my headphones. So I assume you cant use line out? or is the remote a line out also.
    And I assume the remote cable and you headphones act as the aerial, how is the quality? Is it any good indoors, some little FM radios I have used drop to mono indoors. Does this Apple one frop to mono if the signal is weak?
    Someone that has one can perhaps answer
    Cheers
    Ray

    I can at least answer some of the questions.
    The RDS function on this radio/remote is poor in the
    UK. You have to set the radio function on the iPod
    to "Europe" and when you do this RDS info is not
    displayed on almost all radio stations such as
    Virgin, XFM, Capital, GLR, etc. It does display on
    Radio 2 however. At least, this is my finding. So
    it's unpredictable at best. Setting the radio to
    "USA" on the iPod increases the likelihood of getting
    the RDS to display, but the downside is that you can
    then only tune the radio in increments of 2mhz which
    is a bit awkward if your looking for 89.1mhz!
    There are three settings for the radio which you
    choose via the iPod. They are USA/Europe/Japan. No
    setting for Canada, although I'm sure you will still
    be able to use it over there as I can still use mine
    here in the UK even if it's set to USA.
    No, there is no AM radio.
    You can plug headphones into the remote or the
    headphone jack on the iPod, or both at the same time.
    I'm not sure I understand your question about the
    bottom of the iPod. The radio plugs into the iPod
    via the dock connector.
    There is no 'line out' on the iPod. If you mean the
    one on the dock, then no you can't use it as you
    cannot place the iPod into the dock whilst the
    radio/remote is connected as they use the same
    connector.
    Yes the cable of the remote acts as the aerial.
    Reception is surprisingly good where I am, but it
    will of course depend on where you use it. I've
    used it indoors without any drop in signal or it
    dropping to mono.
    Lastly, you need to update the software for both the
    nano or the 5th generation for the radio to work
    (after you do this a radio function appears in the
    menu), and some users have found that doing this
    screws up the iPod. I have mine working on both the
    nano and the 5th gen without any problems.
    There is a review of the radio/remote here. I like
    mine and consider it a worthwhile accessory, but of
    course there will be those who disagree.
    iPod
    Radio/Remote Review.
    Thank you for the reply.
    Think your right USA or Europe will work in Canada.
    Yes I mean the line out via the dock, I just wondered if the remote become a line out socket (I also use a pocket dock for line out for headphone amp as I use a pair of Sennheiser HD600 with my ipod) but of course that is silly becasue you have a volume control on the remote and line out is normaly fixed.
    Based on your views I shall get one in March when in Toronto
    Cheers
    Ray

Maybe you are looking for