Deployment Solution

Does bandwidth throttling from the targetted agent settings affect the outbound communication of tht management platform?

For example:

Given each client including the MP has 512 KB bandwidth throttling.  If there are two package servers, would the actual bandwidth throttling for the package servers be 256KB instead of 512KB?

This depends on your configuration.  If you told package servers in their Targeted Agent Settings policy to use bandwidth throttling, and you added valid throttling periods, and you said they should only throttle when bandwidth is below 512KB/sec, and the package server bandwidth is at 512KB/sec currently, and the throttling unit that you set is a percentage such as 50%, then yes, the package servers will download at 256KB/sec.

That's because you can throttle either based on a specific download rate (e.g. if bandwidth below 512KB/sec, then only use 100KB/sec for the agent) or a percentage download rate (e.g. if bandwidth is below 512KB/sec, then only use 25% of actual bandwidth for the agent; if the computer is measuring bandwidth at 444KB/sec, then it will use 111KB/sec for the agent operations).

This only affects downloads to the agent, not agent communication/policy updates/NSE's/etc.

Does this answer your question?

Each package server will use the settings you target him, no matter if there are other package servers installed, or not.

If you have more than one package server in a remote network, you should have one unconstrained package server (downloading from the NS) and configure the other package server to be constrained (only downloading from the unconstrained package server locally).

http://www.symantec.com/business/support/index?page=content&id=HOWTO45474

http://www.symantec.com/business/support/index?page=content&id=HOWTO44230

Thanks for the info.  It is getting a little clearer. 

In our setup all site servers has a 10% bandwidth throttle.  All sites are remote to the NS.  Our NS is targeted by the targeted agent setting which limits the agents bandwidth.  Does throttling of the notification server's agent affect the bandwidth available to the remote site servers?

Maybe this is a better way to ask.  If the package servers had no throttle applied and the NS was throttled to 512KB all the time, would the package servers be throttled?  Would they have to share the 512KB or would they use all the line?  

The agents are not communicating to each other.

The package server is not influenced by the NS agent settings at all.

It will download the files with its own throttling settings from either the URL or UNC source path.

The agents are affected when downloading packages that they themselves need to execute.  It doesn't affect uploads/serving files/event processing/policy requests/etc.

I'm not sure but I imagine that package servers are indeed affected by the throttle settings when they download packages from the NS so that they can serve them to managed computers in their site.

The NS is targeted by your Site Server Targeted Agents Settings policy, since it is a site server, but again what it's doing is serving files, not downloading them, so the NS isn't going to be affected.

There is no sharing between # of package servers or # of clients.  The clients who have a throttling policy in place ping their package server to determine their download speed.  Based on this result, they either fall within the throttling policy (below threshold) or not (above threshold).

Let's say you create a policy for your workstations where anybody whose test result above is lower than 256KB/sec should use 50% bandwidth.  The workstation receives a task to install Microsoft Office 2010, so it speedtests and the result comes back at 387KB/sec.  As fast as it can, it now downloads Microsoft Office 2010 from the package server.  A throttling policy affecting package server or the NS has no affect on this agent's download speed.

Later in the day, a laptop user with the same policy connects to the network through VPN while at a conference.  It performs the speed test and the result is 122KB/sec.  This falls within the throttling threshold and period, so th edownload will be throttled.  122KB/sec x .5 = 61KB/sec.  The laptop now uses at most 61KB/sec to download the Microsoft Office 2010 package from the package server.

Finally, let's say that the next day, 25 workstations in a computer lab at the remote site are turned on and receive the Microsoft Office 2010 policy.  They speedtest over the WAN and, because it's within the throttling period and it's Monday and people are slamming the network, they fall within the threshold -- each of them speedtested at 200KB/sec.  They will all use up to 100KB/sec to download the package.  Since all 25 will be downloading the package, they use 100KB/sec x 25 = 2500KB/sec over the WAN to download the package.  I use this example to illustrate that bandwidth throttling is to provide adequate service to the client by not hogging their connection.  It's not really meant to ensure quality of service over a WAN/LAN.  That's something a network administrator should handle through their wizardry.

Does this answer your question?  If not, what goal do you have in your environment that you want to accomplish using bandwidth throttling?

As mentioned above, these throttles are determined by the client, so when it pings the PS or NS and finds what is available, it chunks off 10% of that, and goes from there.  There is no way it can know total bandwidth and simply use 10% of that so the values it uses are ... not accurate.

For example.  10 systems using 10% bandwidth each will use about 80% total (10% of 100 + 10% of 90+ 10% of 81...)  it's not very elegant.  Best results, in limited environments, are achieved by doing the math yourself and putting in hard numbers.