Bandwidth reservation


#1

The slider controls on the bandwidth reservation firmware 6.2 are the same for all WAN sources. Is there a way to split these for each internet source? I would like to reserve a portion for managers on one source, but on the other have no reservation on the other source.


#2

Hi Don,

This makes sense. Do you mind to share the used case on this?


#3

We have two internet sources. One is fiber, the other is DSL. As you probably know VoIP over DSL is not good so I have the VoIP forced to fiber as an outbound rule. In spite of QoS and User Group rules giving priority to VoIP, I have found the bandwidth reservation a more reliable way to protect a certain amount of bandwidth for VoIP. I suspect my phone server is not properly tagging the packets for VoIP so the router doesn’t give it priority.

I want to reserve bandwidth for VoIP on the fiber WAN. I listed the phone server as the Manager user group and reserved bandwidth for that group. That instantly improved audio quality. As the available Peplink bandwith reservation options are currently configured I am also reserving bandwidth on the other source, which is a waste due to my outbound rules forcing the fiber link.


#4

Hi Don,

I think your requirement can be achieved after have better understanding on this.

WAN:-

  • WAN1 = 10Mbps
  • WAN2 = 100Mbps

Assumed Bandwidth reservation as below:-

  • Manager = 50%
  • Staff = 25%
  • Guest =25%
  • Only phone server configured under Manager Group.
  • IP for the rest devices put under Staff and Guest Group.

Outbound Policy:-

  • Enforced phone server (under Manager Group) to WAN1.
  • The rest enforced to WAN2

Result:-

  1. Phone server will be guaranteed with 100%(10Mbps) when WAN1 was congested since no others user group using WAN1.
  2. Users under Staff and Guest will be guarantee with 50% (50Mbps) each when WAN2 was congested since user in Manager Group not using WAN2. However, if you load balance phone server’s traffic into WAN1 and WAN2. WAN1 Bandwidth Reservation remains unchanged, WAN2 Bandwidth Reservation will be changed to Manager 50% (50Mbps), Staff 25% (25Mbps) and Guest 25% (25Mbps).

So WAN2’s bandwidth will not wasted. Hope this clear your doubts.


#5

Using your example, my intent is not to totally dedicate WAN1 to the phones. I would like to have:

WAN1 20% to phones
WAN1 80% load balance to everyone
WAN2 100% load balance to everyone

If I reserve 20% to Manager, and manager is not using all that bandwidth at a given time, is that capacity available to others? If so then my current configuration is not wasting 20% of WAN2 because Manager group is not using it.

If there are no clients assigned to a group, is the bandwidth assigned to that group available to others? I am only using Manager for one device. Nobody assigned to Staff. In your example I would want WAN1 20% Manager, with the remaining 80% not reserved to anyone.


#6

Hi Don,

If you assigned 20% for Manager, 80% for Staff (I assume no IP assigned to Guest, so ignore this group for our discussion). Bandwidth Reservation for user groups still applied to all WANs by default. Anyway, the Bandwidth Reservation is scalable based on your Outbound Policy settings. Below is the scenario.

  1. IP under Manager enforced to WAN1 and IPs under Staff enforced to WAN2
  • WAN1 reserved 100% for Manager.
  • WAN2 reserved 100% for Staff.
  1. IP under Manager enforced to WAN1 and IPs under Staff load balance to WAN1 and WAN2 <— Your requirement falled under this scenario
  • WAN1 reserved 20% for Manager and 80% for Staff
  • WAN2 reserved 100% for Staff.
  1. IP under Manager load balance to WAN1 and WAN2, and IPs under Staff enforced to WAN2
  • WAN1 reserved 100% for Manager
  • WAN2 reserved 20% for Manager and 80% for Staff.
  1. IP under Manager load balance to WAN1 and WAN2, and IPs under Staff load balance to WAN1 and WAN2
  • WAN1 reserved 20% for Manager and 80% for Staff.
  • WAN2 reserved 20% for Manager and 80% for Staff.

#7

I have also question about this Bandwidth reservation. I have played with this for a while and confirmed it is really good feature. But please could you explain in more detail how this function determines available bandwidth if one is using only LTE connectivity on moving vehicle? For example, I have four LTE connections in my HD4, which have actually variable bandwidth due to handovers between LTE towers/switching over between cells and operators. Is there any algorithm that checks really available bandwidth in real time?


#8

Hi,

Majority cellular connection doesn’t have a committed bandwidth. So Bandwidth Reservation is not applicable for Cellular connection.


#9

Evening TK
We have the issue regarding fluctuating internet speeds and would like to know if there is a way for a Pep to actively monitor throughput and allocate accordingly
e.g. if 100 Mbps is going through one connection and no packet loss, device knows the internet is great and implements relaxed rules.
Meanwhile, if it was only seeing 1 Mbps going through and lots of packet loss, it would enforce a restricted profile. (would be nice to specify more then two profiles e.g. 100 Mbps +, 20-100 Mbps, 20- 5 Mbps, 5- 1 Mbps and under 1 Mbps)
Thank You


#10

Hi,

Thanks for the suggestion. I am thinking the logic behind. In order to monitor the throughput actively, we need to send out some file regularly to do the speed test. This will cause the bandwidth consumption especially cellular connection always having quota limit. May I know this acceptable and make sense for you?


#11

Hi TK,

I know I a resurrecting an old thread here, but this is very useful for our customer base. Here in Australia, we have some providers with unlimited data sim card plans, meaning that the data used for the speed test is inconsequential. Is this something that is still actively being looked at?


#12

Agreed for unlimited data sim card plan, the usage for the quota is inconsequential. As TK explained also, to know the available bandwidth for the WAN connection, high network traffics need to send to the WAN regularly and this actually congesting the WAN in-order to allow the actual traffics send via the WAN. Especially for cellular WAN, the available bandwidth are inconsistent. The minute you get the estimate bandwidth for the cellular, the next minute the available bandwidth is changed. This actually become a big challenge to use the history data to decide the next action.


#13

Hi Sit,

Thanks for your input, that makes sense. For VoiP deployments, it would be great to have the ability to carve out some bandwidth purely for a voice service (1 or 2 mbits) and allow the remainder of the bandwidth to be allocated for the data network.

Is this currently possible?


#14

I think Application Prioritization (Network > Application) should help instead of Bandwidth Reservation


#15

I have similar problem. Under Application Prioritization we can only associate flags of “High”, “Low” or “Normal” to the categorized packets. What we really need is ability to give some speed quota for every category and user group member. For example “no more than 1,2 Mbps” -> for streaming applications for Management, or vice versa - “reserve 900 kbps for VPN applications for Quest”. The idea of Bandwith Reservation is not suitable for mobile LTE applications - there is really a need for reconsidering this functionality as your HD product line includes LTE as WAN and today it is rather used as primary channel, not as backup. If we would be able to define such kbps-based quotas for applications we could have workaround to this bandwidth reservation that assumes stable channel throughput according to Peplink initial idea.