These little routers are pretty amazing. They aren’t perfect, but they are constantly working on new features and improvements. The community has a huge role in helping steer the engineering team. Every idea I have seen mentioned in the forum is discussed and considered. The best bit is the transparency and accessibility of their team. I have not seen a consumer grade router of this quality with this level of support. I have probably had about 10-15 routers over the last 10 years. I think the Peplink was a bargain for what you get. I really hope in time the company exceeds your expectations.
For what it’s worth, my cheap-o multi wan router also had issues with dropped packets, unresponsive UI (inner iframe wouldn’t load), and frequent necisity to reboot. That coupled with the rubbish outbound routing - I was very happy to get my Balance 30 back🤗
My B30 had WAN port 3 go bad. I was only using 2 WANs, but the warranty was expiring soon - so I sent it in and they sent me another one as soon as they got mine. I think the total time without it was about 7 days. The thing still works, but I replaced it with the balance One Core after adding some AC one Mini AC access points. They are good WAPS for sure. Again, not perfect - but very reliable and consistent. These two are able to be updated as new features are implemented. Oh, and all these new features are available free of charge - that part is pretty cool.
They are already really good on the multi WAN feature, which most routers won’t handle properly. Now they just have to catch up on other stuff, like QoS, IPv6, CLI, etc.
I had the same issues with the TP-Link. The inner frame problem is very annoying, they know about it, but they just can’t fix it.
I hope I don’t get any hardware issue like you did. I live in Brazil, I bought Peplink Balance One in Amazon and redirected it to me using a mail forwading service. It took a month (and customs fees). If I ever have to exchange it, it would take another two months and more fees.
And for some reason I can’t understand, InControl2 says “Warranty Expiry Date: 2017-05-22 (In warranty)”. I bought it in February, installed it this month and warranty expires on May? This is a really shot warranty.
Dang, that stinks about all the customs delays. It sounds like you have other routers around, and the Peplink devices are very reliable - I will cross my fingers that you have good luck. Any piece of electronics will fail in time - you may want to invest in a spare just in case. It is far better to have and not need then to need and not have.
Thanks. I usually leave the old router as a spare when I buy a new one. My backup now is an ASUS RT-AC68U, which I’m using as wireless access point now because it has more range than the Peplink. It’s very reliable using single WAN and it has good QoS, but if I enable Dual WAN all kinds of bugs appears. I don’t think Asus will ever fix this as they don’t even advertise their routers as Dual WAN capable. I’ve sold the TP-Link. There’s also an old Netgear running OpenWRT as a second access point.
It’s not only Youtube uploads. Google Photo Uploader also kills the Internet connection. Basically, any upload that is not limited on the source will kill the Internet connection. I should have taken the pfSense way. Peplink is a disaster.
Do you have multiple WANs? You have the tools to manage your traffic across the links.
Since you are having congestion issues, perhaps you should decrease the configured upload speed to 90% of the capacity for each WAN. Then setup someoutbound policies for port 80 to use the “least used”.
You should also set up your user groups and put some upload bandwith limits to keep one device from consuming your total capacity. I put all of my devices as either guest or staff (managers cannot be throttled). I leave the manager group empty except for a couple of devices (because I dedicate an entire WAN to it during certain times - so I don’t care if it uses all the bandwith).
Since the manager group is empty, reserve 10% of your bandwith for them (that will keep all clients restricted to 90% - which should leave room for your web traffic). You can split the staff and guest at 45% each, or set it how you want. The idea is to reserve enough bandwith for your other traffic.
These settings should really help your situation, but you have to do the work to identify and manage your network traffic. Put an outbound policy for YouTube.com and your phot upload sites - make sure that they are never on the same WAN link, etc.
Now that the technical stuff is out of the way… you are free to have your opinion. You are free to express your opinion. Just a little advice though - don’t dog on a product to a forum that supports that product. You will get better assistance with less badmouthing. The product is really good and once you are familiar with its capabilities, I think you will find that it is one of the best commercial grade routers you can buy. Eventually, Peplink will implement better QOS and possibly some SQM, maybe layer 7 recognition. Right now, their bread and butter is with multi-wan, central web-based management (inControl2), and PepVPN bonding. They do these things very well. It was never advertised as a “lag” or “congestion” killer. Though, with a little effort on your end - it can do a good job at eliminating lag and managing congestion. I truly hope that you eventually start to love your router - it is a good piece of gear. Post back if you have any specific questions on how to best handle traffic scenarios.
I have two WANs:
Fiber: 15 Mbps down / 2 Mbps up
ADSL: 10 Mbps down / 500 Kbps up
I’m already using the algorithm that chooses the lowest latency WAN, which is usually Fiber, but since latency grows up quickly on congestion, it will choose ADSL then.
Upload limits may work so other devices may still navigate, but it won’t help on the device that is generating the load. And it will be always active, because the limit is applied all the time and not only on congestion.
I could put every device on Staff or Guest, and reserve 10% for Manager, but then only Manager will navigate on congestions, other Staff and Guest devices wouldn’t.
I really feel that I’m creating workarounds for problems that shouldn’t exist in any router. I mean, I understand when P2P traffic kills Internet connection, as there are many connections. But I’m talking about a single TCP connection taking the entire link. This shouldn’t happen in any router.
And there is a option on Peplink “DSL/Cable Optimization” that should handle this issue, but apparently it doesn’t.
As for the other stuff. You are right, I know. It’s just that it is a little frustating as I was expecting this router to solve my Dual WAN problems without introducing new ones. If I lived in a decent country it would be easy to buy this router, receive it in less than a week, test it for a few days, return it after I found out about the limitations and get the next one on the list. But unfortunatelly I live in a country where the only Dual WAN router available is that TP-Link aforementioned. To get this Peplink I had to redirect the package, pay expensive shipment, taxes and wait for more than a month.
Have you tried to daisy chain them? I have seen some improvement when running two multi wan routers. It is definitely less straight forward, but it is possible. I think the biggest gains for me is isolating the LAN broadcast chatter and AP management from the WAN functions. Perhaps a combination of the two will provide all the features you want or need - all be it completely managed by your design and implementation. The only gotchas are PNP and multicast domain membership stuff. Imagine a dedicated device for LAN and one for WAN.
I tried that when I had the TP-Link and the Asus, the first for Dual WAN, the second for QoS. It worked, but I had all kinds of problems with the double NAT. UPNP becomes useless in this scenario.
Yep, you found the same that I did. Fwiw - Peplink only supports UPNP on the untagged LAN unless the packet has a TTL higher than 1.
Keep your head up. You are on the right track. You have the best dual wan product you can get. As time goes on - it is only get more feature rich and efficient. That is the beauty of a company that provides firmware updates regularly.
If queue management and congestion control are important to you - look in the feature requests section of the forum. Find a thread that requests that feature and then chime in on it. I don’t know what weight the engineers put on the threads comment count, but chances are that it does play some kind of factor in their decision making process. There are others that want layer 7 QOS, others want coordinated handoffs from WAPS, some are asking for easier very deployments, and probably the most requested is the support for OpenVPN. If you don’t find someone else asking for it - start up a new one.
Take a look at this post by @DocPecos - it may be able to help your situation.
It’s good to know about this UPNP issue. I’m using a VLAN to provide a Guest Wireless SSID. I prefer to use the Asus for WiFi, as it has better range, but for guests I have to use Peplink since Asus guest wireless SSID does not isolate anything when it is on AP mode and it does not support user configured VLANs.
My connection is better now that I lowered the buffer size to 50. It is still slow during large uploads, but now it takes a few instead of several seconds to load a web page. So, BufferBloat is probably the main cause of the issue.
It would be nice to know what Peplink is currently addressing for the next firmware. Queue management would be the non-workaround solution for this issue. I guess it is not that difficult, as the algorithms are freely available.
I already created some feature requests, including better QoS, more groups and more priority levels. Layer 7 QoS is tricky, but more groups and priority levels is probably easy to implement as well, unless they are greedy and decide to make it as an upgrade or a feature available only on expensive models. I’ll create a feature request for queue management too.
considering the default 2000 packets of WAN Connection Buffer Size.
considering a packet usually has up to 1480 bytes.
2000*1480 and we have up to 2,82 MBytes of buffer.
I have 2 Mbps and 500 Kbps of upload bandwidth in my two WAN connections, respectively.
This means I have 12 seconds and 47 seconds of buffer in each of WAN, respectively.
Unless I misunderstood something or I made some wrong calculation, this is way too much, even if we had the best queue management in the world.
Maybe Peplink made this router for connections much faster than mine. For upload speeds of 50 Mbps, which I guess is common on first-world countries, 2000 packets is only half a second of buffer, which seems just fine.
Determine what destination port is being used for the Youtube uploads. While a PC is running the upload, go to Status > Active Sessions > Search and look for the LAN address of the PC. Do this a few times as the port may change, but likely will be in a small range.
Once you have identified the port or range, go to QoS and create a custom rule putting that port range on a low priority
It seems that changing WAN Connection Buffer Size to 40 and 10, respectively, giving each connection only about 250ms of buffer, which is the general rule of thumb, minimized the problem. Now TCP equality works and the upload and requests and ACKs from other connections have the same weight. I still have to see if there isn’t any sides effect though.