Incontrol 'peplink device report'


#1

The traffic reports from in control are confusing to me – my balance 580 is configured with 1 fast ethernet cable, 1 lte modem, and 2 high latency, low-bandwidth satellite links.

The outbound policy is priority based – we use first ethernet, 2nd lte modem, 3rd one satellite system, and 4th the second (more expensive) satellite system. The satellite systems are very low-bandwidth and always congested when in use.

I enabled incontrol and checked the ‘post usage data’ option on the balance 580. Looking at the peplink device report from in control – I’m surprised to see that there is traffic being generated via all WANS at all times … One of my satellite wans (WAN1) has about 180 kbps down and 50 kbps up. The graph drawn by in-control for the in/out of WAN1 shows pretty much straight lines for in and out bandwidth (180 and 50). This is over a period where all the traffic was transitioning over the fast ethernet link … Drilling down by protocol on the incontrol chart shows that all this is attributable to http and https. I’m trying to develop a theory for what’s causing this …

Other relevant details – speedfusion vpn is enabled and also configured with the same priority order as the outbound policy.

Couple of questions:

  1. How much traffic does ‘post usage data’ to inControl typically use? Is traffic always sent via all WANs or does it obey the outbound policy? Could it explain what I’m seeing? If so I might need to disable that because some of my wans are ‘cost per megabyte’ and/or very slow.
  2. If a browser was connected to the remote peplink’s lan https admin port via speedfusion would the incontrol monitor identify that traffic as type (http/https) or as ‘other’?
  3. Any ideas about the traffic profile of the ‘router utility’ iphone app …?

Many thanks,


#2
  1. The InControl traffic is sent via the first WAN that can be used to communicate with the InControl server (report.peplink.com) when it starts up. Attempts are made from WAN1 to WAN5, and then the USB WAN. The protocol is HTTPS (TCP port 443). Data direction is mostly outgoing. The traffic rate is about 1.4MB per hour for the Balance.

Currently there is no way to force the InControl traffic to be routed through which WANs. To minimize the chance the Satellite link being used for InControl traffic, you may move the Satellite link to the last WAN (e.g. WAN 5 in your case).

In considering your situation, we will add an option to allow our users to choose which WAN to use for InControl traffic.

  1. Those web admin traffic over SpeedFusion will be identified as Other traffic.

  2. The Router Utility app uses the same protocol as the web browser to communicate with the router’s web admin interface. But the app just retrieves data and do not retrieve the UI elements from the router. So the app should spend a little less bandwidth than the web browser does for the same pieces of information.

Hope this helps.


#3

Fantastic – thank you very much for this information!

If I can followup with semi-related questions … I’m trying to track down why in-control thinks each of my wans is generating traffic pretty much at their peak load, pretty much all the time … I don’t see this in the ‘real-time’ view on the peplink …

  1. if the priority algorithm is used for outbound policy, along with the priority algorithm for speedfusion (with the same ordering of the constituent wans as used for the direct outbound policy) – is it possible to construct a rough estimate of how much traffic should be generated by a lower priority wan over the course of a day when a higher priority wan is 100% available? Its not clear to me how much traffic is generated by the dnslookup health check policy (with the default failure detection time) and how much traffic is generated to keep the speedfusion tunnel alive …

  2. what’s the traffic profile when ‘manage by incontrol’ is checked but ‘post data’ is unchecked. Is it the same or different?

  3. is there anyway to tweak the router-utility for high-latency links? All the connections seem to timeout too fast for our poor meager over-utilized wans …

Thanks,


#4

FYI, the throughput figures shown on the InControl are the “peak” usages of each 5 min interval. But the Real Time chart on the Balance is showing snapshots of throughput figures. That’s why they look different.

  1. Each health check packet’s size is about 80 bytes. Those packets are sent only when no user traffic is sent out. Below are outbound bandwidth estimations for SpeedFusion on an idle (hot standby) WAN.

“Recommended”: 80 * (60sec / 5sec) = 960 B/min => 57600 B/hr => 1,382,400 B/day => 41,472,000 B/month
"Fast": 8060 = 4800 B/min => 288,000 B/hr => 6,912,000 B/day => 207,360,000 B/month
"Faster": 80
(60/0.4) = 12,000 B/min => 720,000 B/hr => 17,280,000 B/day => 518,400,000 B/month
"Extreme": 80*(60/0.1) = 48,000 B/min => 2,880,000 B/hr => 69,120,000 B/day => 2,073,600,000 B/month

  1. When the “post data” is unchecked, the in and out bandwidth usages are also 82 bytes per 25 secs. I.e. 82 /25 * 60 * 60 = 11808 B/hr => 283392 B /day => 8785152 B/month for one direction.

  2. Currently the timeout value is 10 secs which should be long enough for most cases. If possible, could you send us a video of how the timeout occurs?