Mikrotik. Or, you can take an older PC, add two or more network cards and install pfsense or just Linux and use that as a router.
3 Likes
If you store something for a day, you get 1/30 of the monthâs payout for it.
Thank you. So if its there for part of a day, you would get 1/30th? Thinking of data with very short expiry.
CC
It has already been answered in this very thread, see here:
4 Likes
Well, then, youâve invested in your own router. Thatâs the point of the argument, people would prefer not to, nothing more to it.
If I want to fill a 20TB drive, I need to buy a 20TB drive first. Likewise if I want to sustain my max connection speed, I need a router (technically I donât, but thatâs another story) that can sustain it. I donât get the counter-argument. If something doesnât work, you replace it, not complain why itâs not working. This âbut itâs been fine for years!â statement is pointless.
We arenât talking about SNOs making $10/month, we are talking well into the $500/month range. If they canât buy a router with that within a reasonable time (ie less than a year) then they are doing something extremely wrong.
3 Likes
âThey told me to âuse what I haveâ⌠but what I have kinda sucks. Devs do something!â
2 Likes
That has always been a part of Storj. âUse what you haveâ - and then publishes the requirements for uptime that pretty much mean you need a UPS, generator and a second internet connection (later those requirements were relaxed).
Now itâs âuse what you haveâ, but apparently, what some people have is not good enough 
Oh, also, âuse what would be online 24/7 anywayâ and then proceeds to say that:
- Running a node in a VM is bad
- You should keep the node on a separate vlan (for your own security).
Add those two together and it means dedicated hardware, not âwhat would be online anywayâ
I run my node in a VM because I use that host to run other VMs.
Just because I use an old eeepc 701 connected to a 20TB usb drive because âI had that alreadyâ doesnât invalidate the guidelines. The same way that machine would be overloaded if I put it into a Gbps situation (assuming that it I could even get a Gbps card running with it), is the same situation with this test data: You have it, but if it falls over under this load, itâs not good for the network overall. You can try nursing it to its full, but you can also get something that actually works without any issues and opens up the way for expansion down the road.
No, I havenât. Iâve always had this setup and it doesnât help Storj at all. Since itâs a dual NAT setup (yes, I know) both routers have to deal with the full load same as if they were used on their own. And both routers cope just fine. The only reason I have the second router is because my ISPs wifi sucks and has way too few ports. So Iâm routing everything except TV/phone through the Asus router. I didnât buy it for Storj and if anything the dual NAT should make performance worse, not better. Though Iâve not seen a negative impact enough to make me change my setup.
Get your ISP to bridge their router and it will be like itâs not even there. ISP stuff (VOIP/TV) gets handled on separate VLANs anyway, so they are only bridging the internet traffic to your downstream (secondary) router.
1 Like
Canât. VOIP credentials are stored on the provided router and they refuse to provide those to use anything else. Trust me, Iâve tried better solutions. ISPs suck. That was my conclusion. Still they give me a static IP (unofficial, but it hasnât changed in over a decade) and super stable and consistent 1gbit. And this works just fine. I doubt it would make much of a difference.
You donât need the credentials. They will be bridging the internet VLAN to (usually) port 1: Itâs an interface to interface bridge. By having their VOIP on a different interface (vlan), that means the login to the VOIP server is done within that provided router and you donât need to do anything on your end.
Some L1 support may not know how to do that, ask to escalate and it can be done.
If VOIP ant internet are on separate vlans, then you can insert a managed switch between the ISP and the router, then configure the switch to pass the VOIP vlan to the ISP-provided router and pass the internet vlan to your router.
This may be possible to do without the ISP being involved, depending on how closely they monitor this.
You missed the part where the switchâs interface doesnât match the coaxial/dsl line coming in.
Coax would be a problem, dsl can probably be made to work by getting a dsl modem (that can be configured in bridge mode), then connecting the switch to it. GPON may also be a problem, depending on how the ISP identifies the device.
Iâm just used to ISPs providing Ethernet (from a switch in the building, fiber to copper media converter or GPON ONU) and a separate router.
ONT, but yea, depends on the interface coming in. Most ISPs lock their âoutâ side to their routerâs mac, so you canât add whatever you want in front of it.
Cloning the MAC is the easy part.
1 Like
Go do that. Then watch as everything crumbles as your switch now sees duplicate macs on two different interfaces.
You get into the Mac address issues as well then (yes I know you can clone those on many devices). They did a lot to lock things down. But letâs not derail the thread with this tangent. I appreciate that you all want to help, but Iâve given up on this and it doesnât cause an issue for me. My point was simply that both routers had no problem and my 1gbit got saturated during some of the tests. So consumer routers can definitely work as long as they arenât the most crappy old stuff. Even when using 6 IPâs.
1 Like
Hardware that can sustain it untill now is different of what we need today with this ânew patternâ. Again, we are not talking about 1/10 nodes! We are talking about PB+ of data.
Scaling job is not easy⌠if you have infinite money you are going to buy something that can work on 1gbit per node but if you have to manage 100 nodes or more you need to find a compromise. Today pattern broke that compromise
1 Like