Bandwidth detection causes bandwidth consumption to spike

Hi everyone,
We are having a little problem implementing bandwidth detection with FMS, for some reason when enable it, bandwidth consumption spikes significantly I am wondering why? and if there a white paper somewhere that would talk about how properly implement this feature.
Kind regards.
+LA

Hi,
I dont think this should happen anyways can you tell exactly how did you concluded this. I mean to say how are you finding bandwidth usage as well as bandwidth detection.
The following link might be helpful for you.
http://livedocs.adobe.com/flashmediaserver/3.0/hpdocs/help.html?content=00000072.html#1068 463
Regards,
Amit

Similar Messages

  • Bandwidth Consumption - How to Identify

    I am having an issue reconciling what I am seeing on my box and what my ISP is reporting that I am using for bandwidth consumption.
    For a sample hour, they report the following usage:
       Usage: 0.240 GiB In / 0.374 GiB Out
    Which if I am reading this correctly is 240MB in and 374 MB out.
    However, when I use a utlility called vnstat I am getting:
        112 MB In / 247 MB Out
    Question: what other tools should I be trying to use to identify the discrepency.  I thought I could be hacked, but I'm pretty sure I have not been compromised (not 100% though).  I run very few services (ssh / http / smtp / pop3 / imap), run osiris on it nightly and various rootkit detection scripts.
    The CPU utilization is about what I expect; I just don't understand the wide differences in the bandwidth consumption.
    I realize this is a pretty broad question, even a pointer where I can learn more about how to analyze network issues would be appreciated.
    Thanks,
    Chris....

    ok - I'm official outside my comfort zone.  I took a tcp dump and analyzed in wireshark (the sniffer previously known as ethereal).  I am seeing a lot of ARP packets that originzate from my ISP's routers.
    Couple of questions for those who know more.
    1. What is a lot of ARP traffic?  I am seeing 220K packets & 15MB over a 10 minute span.  I would think this is a bit excessive.  Is it realated to the size of their network.  At my last provider, I saw total inbound traffic during quite periods (late at night) of around 10MB per _hour_, so that is my limited comparision point.
    2. Can I safely block this traffic in iptables?  Some of it if it is not directed at my MAC address?  Or should I just not worry about it and filter it out of my network reports when looking for problems.
    Thanks,
    Chris....

  • Network top talker! Bandwidth consumption!

    Hello,
    what is the best solution in cisco world for finding the top talkers in the network.
    I need to know what bandwidth consumption per IP.
    I found out that Cisco IOS router with solarwind using SNMP.
    Does cisco ASA generate this kind of reports?
    Regards,
    Viorel

    just use google and viola! https://supportforums.cisco.com/docs/DOC-6113   netflow stat in ASA.

  • RMI bandwidth consumption

    we're about to deploy a new third party RMI-based application.
    The application is supposed to receive client connection through WAN .
    running an initial bandwidth test to try to detect optimal bandwidth, we assumed worst-case-scenario where a 128kbps
    leased line was used for client connections.
    With only one client connecting, the bandwidth was fully used for 70% of the time and the client's download rate overcame its upload's.
    since we have the sources and license to edit of the application, i looked up tis code a bit and i noticed that most of classes the client had loaded and was initiating the calls from existed only at the client side.
    if those classes,existed in the server's classpath as well would that diminish the client's upload rate assuming that then the server would have whatever objects definitions the client is referencing through its calls ?
    I might then alter the code in a way to redeploy the classes efficiently.
    i don't know if i was clear enough, i could use some advice though.
    thanks

    ejp wrote:
    The server only needs the classes that are reachable from the remote interface(s), and it most probably already has them. RMI doesn't send classes over the wire unless the java.rmi.server.codebase feature is enabled, and in any case it is a side-channel, usually via an HTTP server.i read the dynamic code downloading using RMI ref. document but one thing i still don't understand
    "+if the remote method receives an object instance, for which the remote object cannot find the class definition locally in its CLASSPATH.he class of the object sent by the client will be a subtype of the declared parameter type. A subtype is either:+
    +* An implementation of the interface that is declared as the method parameter (or return) type+
    +* A subclass of the class that is declared as the method parameter (or return) type+
    how can this case occur, and if it does the remote object will look for the class's definition at the codebase of the client , am i right?
    The bandwidth utilization of an RMI invocation is a more or less linear function of the size of the arguments and return value, including all non-static non-transient objects reachable from them.if i understand you well , to minimize the consumption, the objects that are to be used as arguments you are talking about should be coded in a way that uses transient and static wherever applicable and possible,and the smaller objects you provide the less bandwidth you use.
    am i right ?
    Other than that is there any way or any measures that can help in this direction?
    regards

  • Bandwidth Consumption through RV180

    Hi,
    I am working with a customer utilizing a RV180 as their EDGE device and an SF102 small business 24 port switch for switching.  At times, we are experiencing issues whereby some device or devices are consuming all available WAN bandwidth, causing the internet to be poor-performing at best during this timeframe.  We are looking for a way to determine what specific host or hosts are consuming all of this bandwidth.  I know with an ASA you can set up Netflow through PRTG - can this be done on the RV180 and if so, how can I set this up?  Or is there some other methodology I can use to find the culprit host(s)?
    Thanks,
    Mike

    ok - I'm official outside my comfort zone.  I took a tcp dump and analyzed in wireshark (the sniffer previously known as ethereal).  I am seeing a lot of ARP packets that originzate from my ISP's routers.
    Couple of questions for those who know more.
    1. What is a lot of ARP traffic?  I am seeing 220K packets & 15MB over a 10 minute span.  I would think this is a bit excessive.  Is it realated to the size of their network.  At my last provider, I saw total inbound traffic during quite periods (late at night) of around 10MB per _hour_, so that is my limited comparision point.
    2. Can I safely block this traffic in iptables?  Some of it if it is not directed at my MAC address?  Or should I just not worry about it and filter it out of my network reports when looking for problems.
    Thanks,
    Chris....

  • 1250AP's with 4400 Controllers - LWAPP WAN Bandwidth consumption

    Does anyone know and, preferably, have a link on how much bandwidth a Lightweight AP consumes when deployed across a WAN link? I know Cisco keeps saying no more then 8 across a WAN managed by a centralized controller, but I can't find anything specific. Thanks so much in advance.

    There's a lot of components in LWAPP packets which will vary according to different situation.check this for LWAPP detail: http://tools.ietf.org/html/draft-ohara-capwap-lwapp-02
    actually, A discovery request packet is 97 bytes (including the 4 byte FCS),A discovery response packet is 106 bytes (including the 4 byte FCS), Packet sizes for the join request and response messages will vary based on the MTU supported by the transport(1596 or 1596+1500),The initial exchange between the AP and the WLC (ap-manager interface) is approximately 6,000 bytes and a one-time configuration change averages 360 bytes and involves 2 packets each from the AP and the
    WLC's ap-manager interface.An RRM-related information exchange takes place once the AP has been provisioned. A typical exchange
    between the AP and the WLC (ap-manager interface) is approximately 1400 bytes. In the event of an RRM-related configuration change, there is a four-packet exchange between the AP and the WLC's apmanager
    interface. This exchange averages 375 bytes.The system heartbeat, coupled with fallback mechanism, is 4 packets every 30
    seconds and comprises of the following packets:
    LWAPP ECHO_REQUEST from AP (78 bytes)
    LWAPP Echo-Response to AP (64 bytes)
    LWAPP PRIMARY_DISCOVERY_REQ from AP (93 bytes)
    LWAPP Primary Discovery-Response to AP (97 bytes).
    There are two ongoing RRM exchanges. The first one, at every 60-second interval is the load and signal
    measurement and consists of 4 packets. This exchange always adds up to 396 bytes, as following:
    LWAPP RRM_DATA_REQ from AP (107 bytes)
    LWAPP Airewave-Director-Data Response to AP (64 bytes)
    LWAPP RRM_DATA_REQ from AP (161 bytes)
    LWAPP Airewave-Director-Data Response to AP (64 bytes)
    The second sequence of packets is the noise measurement (including a statistics information request and
    response sequence) done every 180 seconds. This is a short (0.01 seconds typically) exchange of packets
    and averages 2,660 bytes approximately. It consists of the following packets:
    LWAPP RRM_DATA_REQ from AP
    LWAPP Airewave-Director-Data Response to AP
    LWAPP RRM_DATA_REQ from AP
    LWAPP Airewave-Director-Data Response to AP
    LWAPP RRM_DATA_REQ from AP
    LWAPP Airewave-Director-Data Response to AP
    LWAPP RRM_DATA_REQ from AP
    LWAPP Airewave-Director-Data Response to AP
    LWAPP STATISTICS_INFO from AP
    LWAPP Statistics-Info Response to AP
    LWAPP RRM_DATA_REQ from AP
    LWAPP Airewave-Director-Data Response to AP
    LWAPP RRM_DATA_REQ from AP
    LWAPP Airewave-Director-Data Response to AP
    LWAPP RRM_DATA_REQ from AP 00:14:1b:59:41:80
    LWAPP Airewave-Director-Data Response to AP
    LWAPP RRM_DATA_REQ from AP
    LWAPP Airewave-Director-Data Response to AP
    LWAPP STATISTICS_INFO from AP
    LWAPP Statistics-Info Response to AP.
    Rogue measurements are done as a part of the scanning mechanism (for more details, refer to the Auto RF
    and Rogue Detection whitepapers) and included in the above RRM exchange every 180 seconds.
    The LWAPP data frame header adds 6 bytes to the existing 802.11 packets.
    Since LWAPP frames can be fragmented, a Fragment ID field is included and the total packet size can be
    determined by adding the original frame and the IP Fragment that follows (It is important to note that the IP
    Fragment that follows is not encapsulated in any LWAPP headers).
    In conclusion, operation of LWAPP does not introduce heavy bandwidth requirements on the infrastructure and in most typical deployments, there would be no such need to add extra capacity to the infrastructure to accommodate Cisco's Unified Wireless Architecture.

  • ACPI.sys, NDIS.sys, and TCPIP.sys causing massive DPC latency spikes

    I am running Win 7 Ultimate. I recently reinstalled my OS. Using LatencyMon and DPCLatency checker, I've noticed that these three drivers are causing massive spikes. How do I replace these files with non-problematic files?
    (Originally, my computer was on Home Prem, I did an anytime upgrade to Ultimate, this time I did a clean install)
    Thanks.

    Dear all,
    I am also doing home recording with my computer, via firewire and I also see latency spikes.
    Its for an advanced home user difficult to judge, what is causing issues.
    Audio clicks and drops were mainly caused on my (high performance system) by non optimal combination of Firewire Card/Chipset, Windows 7 Firewire driver and driver of the audio Interface, but which I could finally solve.
    My Solution for my system:
    - 1st I had to buy a new firewire card with recommended chipset of audio interface vendor
    - Additionally I had to use the Win7 Firewire "legacy" driver
    -> Then my audio clicks and drops went away.
    Month later the vendor releases a new version of his Mixer/driver for the audio interface (Saffire mix control).
    And what shall I tell:
    - the normal Win7 firewire driver could be chosen (which seems to be more optimized in terms of latency)
    - under Cubase (with setting "optimize cubase for windows") the firewire latency went down from 200us to even 76-90us on average.
    Of course Microsoft shall work on optimizing their OS, I want an optimized OS as good base for everything.
    But all my problems went away in terms of recording, at least on my system, when I optimized things around Mircrosoft.
    With a Laptop this might be a different issue ..
    LatencyMon v3.0: CONCLUSION
    Your system seems to have difficulty handling real-time audio and other tasks. You may experience drop outs, clicks or pops due to buffer underruns. One or more DPC routines belonging to a driver running in your system appear to be executing for too long. At
    least one detected problem appears to be network related. You may have to disable wireless adapters to get better results.
    OS version:                                          
    Windows 7 Service Pack 1, 6.1, build: 7601
    CPU:                                                 
    GenuineIntel Intel(R) Core(TM) i7 CPU         860  @ 2.80GHz
    Logical processors:                                   8
    RAM:                                                 
    17170518016 total
    VM:                                                  
    2147352576, free: 1916923904
    Reported CPU speed:                                   2807 MHz
    Measured CPU speed:                                   2722 MHz
    Highest measured kernel timer latency (µs):           1289.9413
    Highest measured SMI routine or CPU stall (µs)        98.8614
    Highest DPC routine execution time (µs):              1978.9440
    Driver with highest DPC routine execution time:       ndis.sys - NDIS 6.20-Treiber, Microsoft Corporation
    Highest reported total DPC routine time (%):          0.9446
    Driver with highest DPC total execution time:         Wdf01000.sys - Kernelmodustreiber-Frameworklaufzeit, Microsoft Corporation
    Total time spent in DPCs (%)                          1.7380
    DPC count (execution time <250 µs):                   3443062
    DPC count (execution time 250-500 µs):                0
    DPC count (execution time 500-999 µs):                68
    DPC count (execution time 1000-1999 µs):              6
    DPC count (execution time 2000-3999 µs):              0
    DPC count (execution time >=4000 µs):                 0

  • Trigger invalidations, how to detect cause?

    Hello,
    is it possible to somehow detect what causes object(PKG/PRC/FNC/TRG) invalidations?
    Application users doesn't have righst to compile the object. After invalidation object can be again successfuly recompiled by ALTER XXX COMPILE; command. For example we are using after logon trigger for application users, this trigger gets sometimes invalidated for no obvious reasons, users are unable to login to the database because they're hitting permission denied upon trying to compile this trigger. Any hints on how to detect what is the cause of these invalidations?
    Thank you,
    D.

    Hm, what if I found out that last DDL time with the latest date were the following objects:
    SYS.DBMS_STANDARD
    SYS.DICTIONARY_OBJ_NAME
    SYS.DICTIONARY_OBJ_TYPE
    SYS.STANDARD

  • CS5: What could be causing my CPU to spike?

    Machine Specs:
    Mac OSX 10.6.6
    2.66GHz Quad Core Intel Xeon Processor
    8GB RAM
    Running Adobe CS5 Design Standard
    Background story:
    I have inherited a magazine from a co-worker. The magazine has been worked on for years by our business and has undergone some changes, but kept its core design/layout. From my observations, before each issue the magazine is stripped and resaved to keep the content and advertising that remains constant. There's a large collection of swatches and styles, some that are used, some that are not. Some ads are created inside the document, but most are placed PDFs. Even some vector logos are imported directly to the document through copy/paste. In the past, the document has been having some issues (crashing, trouble opening) on my co-worker's computer, but has been workable. (He has a 3.3GHz processor, if that helps.)
    My problem:
    I tried opening the document, and immediately the CPU usage in the my activity monitor jumped to 100%. There's around 5GB of free memory, but the CPU spikes. Indesign will go into 'Not Responding' and not do a thing. If I force quit and re-open the application, the file I intended to open is there as if it was recovered from a program crash. I have tried opening an IDML file created of the same document to ecounter the same problem. Thirdly, I created a new document at the same specs, and imported all styles/swatches and moved the pages fro mthe original document. Same problem once I closed and tried to reopen the document. It's obvious that there is something lurking in this file which is causing problems.
    My possible Solutions:
    1. Find the problem and eliminate it. This would obviously be best, because there is quite a lot of things in this document to keep.
    2. Make do with having to force quit Indesign and reopen it to open the file. This idea makes me a little uncomfortable, but I'm sure that saving before closing would make sure nothing is lost in the recovery, right?
    3. Create a new document and layout from scratch, using only required swatches and styles and copy/pasting elements from the other document when required. This may take a little more time than other issues, but could be a healthy step forward.
    Regardless of the solution, it has been interesting (and rage-inducing) trying to find what the possible problem could be. Does anyone have any ideas, or has anyone encountered the same problem?

    What you have right now is a disaster waiting to happen. It's probably impossible to say with any certainly where the problem is at this point. Repeatedly saving the same document issue after issue allows little corruptions to build up and multiply until one time it will simply fail altogether.
    It it were my job, I'd open the file as a copy, then IMMEDIATELY export to .idml. If there are no CS5 -only features in use I'd open that in CS4 and export again, this time to .inx. Open that, or the .idml if you can't make the .inx, in CS5 and save as a reference file with a new name. Hopefully any corrupt styles have been cleaned up. Next I'd start a brand new file with the same page size. If there aren't a huge number to do, I'd recreate all the master pages from scratch, but if that's not practical you can add them in the next step.
    Add the new file and the reference file to a Book file, and set the reference as the synch source. Add Master pages to the list of things to synchronize if you didn't make them by hand, then make sure everything else is checked and synch the book, but don't do any numbering updates. Save the new file as a TEMPLATE. Add anything to the template that will remain the same over multiple issues, and start each new issue going forward by opeing the template and saving as the new issue. Stop recycling the old issue over and over.
    You can add ads to the template, but I wouldn't since they really arn't the sort of thing that is "static." To speed things up you can make a library or a folder of snippets for them and drag them into each new issue as necessary. This won't really take any longer, once it's set up, than the time it takes to remove unwanted content now. By using the template you are always starting with a fresh second-generaztion copy, which hopefully is clean. If you still find you are having trouble, you may have to rebuild all of your styles from scratch as well. and replace the ones you imported.

  • What's causing periodic load average spikes?

    Hi all. I have a quick question, which should be easy to solve, but I haven't been able to yet.
    Approximately every 3.2 hours (the period is very regular, and a bit more than 3 hours), the load average on my XServe climbs up to 1.0, stays there for a few minutes, then falls back down to its average level (normally less than 0.1). (I've graphed it out which is why I know it's so regular). But I haven't for the life of me been able to figure out what's doing this. When it spikes up like that, 'top' still shows a relatively idle CPU (~95% idle), 'ps alx' shows only the 'ps' command itself having an 'R' STAT flag, and 'sudo fs_usage' shows periodic bits of activity, but nothing crazy. How can I find out what's going on? I've also tried commands like 'iostat 2' and 'netstat 2' to see if anything obvious is happening, but all seems relatively quiet... other than the load average as reported by top, uptime, etc. I know that a load average of 1.0 isn't all that high and the server still feels responsive during these spikes, but it should be mostly idling closer to 0.0. I want to make sure that if something is starting to bog down the server that I nip it in the bud before it starts to become a real issue.
    I know, I could just reboot the machine and see if the problem goes away. But there are two main reasons I haven't done so. First, I really want to know what's happening, and even if rebooting the machine does fix the problem, I won't have learned anything from it and there would be nothing to prevent this from happening again. And secondly, I had a bad experience once where rebooting the machine remotely didn't run smoothly - for some reason, the primary hard drive didn't spin up quick enough to be recognized, and it booted into its secondary recovery hard drive, which didn't enable me any kind of remote access to the machine... so it meant several hours of downtime while I drove to the XServe's location to manually restart it again. The box's availability is critical to a business' operations, so I don't want to risk the possibility of downtime unless all else fails.
    Any advice on what might be causing this, or what I can use to figure out what's contributing to the load average during these times?
    Thanks!
    Dan

    Just to add a bit more food for thought for those experts out there...
    A few days ago, I tried changing the syslogd configuration to log debug information to see if something could be tracked down. After changing the syslogd.conf file, I sent a HUP signal to the syslogd process to try and get it to re-read it. I must've done something wrong, since I never did get debug information logging. But something else happened which was interesting.
    I got a discontinuity in my regular 3 hour load average peaks. The next one happened an hour or so after I sent the signal (maybe 2.5 hours after the previous peak) and then continued on with a 3.2 hour periodicity. Then, the peaks ramped down and I thought the problem may have settled down. But then, they came back, and this time it doesn't quiet down to a load average of 0 in between peaks - it's around 0.25 now. Before, the server was normally at a load average around 0, peaking to 1.0 every 3.2 hours for maybe 5 minutes. Now, it sits normally at a load average around 0.25, peaking to 1.25 every 3.2 hours for 5 minutes.
    I don't know if the HUP signal to syslogd changed anything, or if it was just coincidence.
    Does this bit of info suggest anything else I could try?
    Thanks!
    Dan

  • Loops detected causing slowness on entire Vlan?

    Hello Cisco Community! I'm hoping someone out there can confirm my diagnosis. Still a little green when it comes to this kind of thing.
    Issue
    Intermittent connectivity to Vlan105 from all other Vlans and slowness between hosts in Vlan105. 
    Discovery
    STP detected a loop for a MAC among two ports of a single Oracle Database Appliance (ODA). Turns out the DBA had kicked off a clone job from the test ODA on another Vlan to the ODA in Vlan105. Doing this saturated the firewall interface for Vlan105. 
    Diagnosis
    Obviously something is not configured correctly for the ODA configuration on Vlan105 which will be looked at and remedied. The STP loop detected should not have caused the intermittent connectivity and/or slowness within and two Vlan105, correct? 
    I believe we have two things going on here. The network configuration and the pure fact we saturated the interface. 
    Any ideas on this would be greatly appreciated. I'm if for the network team here and this is the best we could come up with among myself and system administrators involved in this.

    Prod N5k2 is actually the secondary of a VPC configuration but during this issue Prod N5k1 was not in play due to a bad bootflash. The firewall and ODA servers are not necessarily plugged into the same FEXes but for this purpose I didn't think it mattered. There are no other switches in either environment to be the root bridge other than either N5k in the VPC pair. The firewall does all inter-vlan routing for the prod network.

  • Why does the new version of iTunes cause my CPU to spike?

    I've downloaded the newest version of iTunes (and even reinstalled it, to be sure).  [Version 10.7.0.21].
    When it is idle, or at least, when it is supposed to be idle, it causes my CPUs to run up quite a load.  Up to 27% from 4%.  This appears to be centered on the program AppleMobileDeviceService.exe, which accounts for about 25% of my overall processor load.
    The processors cool off and quiet down as soon as I end that task. 
    Is this a known bug in the new app, or is this just me?
    -Steve

    I just found out from another post on how you can fix the issue.  Open QuickTime Player on your computer and go to Edit >> Preferences >> QuickTime Preferences.  Once your in QuickTime Preferences, click on the "Advanced" tab then uncheck the box that says "Enable Direct3D video acceleration".  I tried this and it seemed to solve my problem.  Hope it works for you! 

  • X-Fi: No sound at startup: Problem detected, cause still unkno

    After a minucious analysis of my problem with the SB X-Fi(see my previous posts), I think I have pinpointed it.
    Problem: There's no sound at startup, even if everything looks fine. Suddenly, sounds comes (sooner or later), and it stays, even after rebooting. Next day, there's no sound again.
    My system info:
    Motherboard: Asrock Conroe 945G-DVI
    Memory: Kingston DD2 Gb
    Hard Disk: Western Digital SAT2 200JS
    I have some enigmatic things in Device Administrator > Audio, Video and Game Devices:
    - Audio Codecs
    - Video Codecs
    - Inherited Audio Drivers
    - Inherited Video Capture Drivers
    - Multimedia Control Drivers
    - Creative SB X-Fi
    I don't know that they are, whether I should delete them or not... there are no exclamation marks, but the X-Fi works erratically.
    Any ideas or suggestions, please? The system is brand-new, and I'm completely baffled.

    I took my computer to the retailer. After fi've days, they sent it back to me with a note saying : After fi've days of testing, no error found. No problems found.
    Cool. First day, I had sound at startup. Everything was nice. Yesterday, I had sound at startup. Nice. Today, suddenly, I no longer have sound.
    I'm completely baffled. Could it be the speakers? I have the 2700 Inspire Speakers by Creative, and when I was installing the X-Fi card there was a message about this particular set, but I didn't read it, and I'm unable to find it again.
    Can ANYONE please suggest anything to me? I don't know what to do.

  • Monitor the consumption of bandwidth

    Hi,
    I Need help to choose the best solution to monitor the consumption of bandwidth, particularly who are consuming the bandwidth. With the SNMP protocol I can know the consumption of bandwidth but I can't know who are consuming. What are the alternatives?
    Thanks to all

    If the L3 3750 switch is the gateway for your hosts, it's not very easy to capture flow-level data. Netflow is generally only supported on routers and high end switches like the Catalyst 6k. (Though there is a feature module for the newer 3750X's).
    For an ad hoc look at the data, one could span a port and parse the data in a protocol analyzer. That's generally overkill for a simple look at flow patterns though. RMON probes are also possible thought they have generally fallen into disuse in recent years.
    Do you really want to look at LOCAL bandwidth consumption (i.e., among ports on a LAN switch)? Folks seldom do that at any level of granularity except in troubleshooting specific problems since LAN bandwidth typically exceeds WAN by an order of magnitude or more and is generally over-provisioned. It's usually much easier to just watch, say, switch uplink utilization in the aggregate and add additional ones (or move servers around logically) if utilization becomes an issue.
    Seeing the overall architecture and components of your system would allow us to give a better assessment.

  • Strange network bandwidth management

    Hi,
    I've noticed some strange behaviour of Remote Desktop Services.
    I have remote Windows Server 2012 R2 running in data center. There is 100 Mb/s download / 100 Mb/s upload internet link tested with speedtest.net. I have also two laptops with Windows 8.1 in my office running on 25 Mb/s donwload / 3 Mb/s upload link. One
    of the laptops has 1280 x 800 screen resolution and the second one has 2880 x 1620 pixels.
    I tested the connection speed by downloading 2 GB file from Windows Server to my laptop. File was downloaded with 25 Mb/s speed as expected. So the connection is perfectly OK.
    I've configured our application as RemoteApp on Windows Server and have been testing this environment on my laptops. Some modules of our application are "rendering intensive" - they change some big portion of screen 10 times per second.  And
    here the problem starts.
    When I test RemoteApp only on my low resolution laptop all work great. RDP protocol utilizes about 1 Mb/s of bandwidth.
    When I test application on my high resolution laptop RDP protocol utilizes about 5 Mb/s of bandwidth and application presents very low frame rate (1 frame per second maybe). I don't know why this happens. I have 20 Mb/s link and only 5 Mb/s is being used
    by RDP.
    And magic happens...
    When I start my RemoteApp on both laptops frame rate is perfect on both of them! Windows Server is uploading 14 Mb/s (1 Mb/s to low resolution laptop and 13 Mb/s to high resolution one). It streams 6 Mb/s more to high resolution laptop than in "single
    laptop" scenario! When I close or even minimize (!) application on low resolution laptop the frame rate on second laptop degrades. When the application is resumed, frame rate goes up.
    The same behaviour is observed when I run RemoteApp on high resolution laptop only but I connect through RDP to my Windows Server and turn on session shadowing. I observe my laptop's session through shadowing and frame rate magically goes up.
    It's strange that streaming RDP sessions to two laptops produces better experience than doing it only for one client. I'm able to reproduce this behaviour every day, every time in two locations with different internet service providers.
    I turned off all magic fair sharing (disk, CPU, network) on Windows Server but without any success. The same about turning off UDP connections or connecting with and without RD Gateway. I think that there is some magic algorithm for resources management
    that is causing it. I've been searching internet and testing different scenarios for week but without any luck.
    Have anybody seen something like this? I would be very grateful for your help.
    Wojtek Małota

    Hi Wojciech,
    Bandwidth consumption when using RemoteApp programs could increase and decrease, depending on the application and its usage, when compared to using a remote desktop. There are lot of factors which can be taken into considerations when measuring performance
    for RemoteApp\Desktop. You can go through below article for detail as also there is RDP performance whitepaper to understand clearly.
    RDS: Choosing between Delivering RemoteApps versus Full Desktops
    http://social.technet.microsoft.com/wiki/contents/articles/1366.rds-choosing-between-delivering-remoteapps-versus-full-desktops.aspx
    Hope it helps!
    Thanks.
    Dharmesh Solanki
    Please remember to mark the replies as answers if they help and unmark them if they provide no help. If you have feedback for TechNet Support, contact [email protected]

Maybe you are looking for