
Nexland Pro800Turbo Load Balancing Router Review 141
An anonymous submitter writes "Found this review today over at OverclockersClub.com. Apparently this router can load balance two broadband connections like DSL, Cable, or T1. The router can also act as a backup feature in case one of the broadband connections goes down, the router will automatically switch to the connection still working." At $400, it's not gruesomely expensive either, and I guess if you're willing to pay for two broadband connections anyway... The spec sheet (PDF) has more information.
why not a software solution? (Score:3, Informative)
Super Sparrow is a distributed load balancing package also by Horms (formerly of VA Research|Linux|Software|Spacecraft|Doohickeys) that uses BGP route information to decide which server ought to service a request. Neat stuff. Super Sparrow is not ready for deployment, and appears to be on a back burner (due to VA's disinterest in such things these days, probably).
LVS is the project to beat in this space, by a long ways. It is very very solid, and extremely efficient. Wensong is quite an impressive nerd.
Re:why not a software solution? (Score:2, Troll)
A machine running load balancing software is still hardware that must be managed. I like a small router with no moving parts over something with a hard drive that makes noise and heat.
Re:why not a software solution? (Score:3, Informative)
Re:why not a software solution? (Score:1, Troll)
Re:why not a software solution? (Score:1)
I'd be real surprised if the Verizon DSL available in my area shared a gateway with the Charter Pipeline cablemodem service.
Re:why not a software solution? (Score:1, Insightful)
Re:why not a software solution? (Score:2, Informative)
Big difference. You can't run a load balanced web site with a device that works in this direction. To do that, you need a big pipe in that gets load balanced to *your* servers.
Re:why not a software solution? (Score:2, Interesting)
Re:why not a software solution? (Score:1)
Re:why not a software solution? (Score:1)
In the case of network where even older CPUs can do packet-fu with enough spare computrons to simulate a nuclear explosion (did I say in a timely manner or high detail or anything?) I think that a hardware based solution might not have the edge you're assuming.
Hardware kicks ass for repetitive fixed functions (screw flying cars, where is my hardware constructive solid geometry raytracer), but normal CPUs are there and programmed for the task already and aren't strained by it much.
Re:why not a software solution? (Score:2, Interesting)
Re:why not a software solution? (Score:2)
Sure but then you have three pieces of hardware where this could be done with only the one mentioned. That increases probability of failure, admin overhead, power consumption and heat generation. Not to mention that the hardware costs would be more for the quid based solution depending on existing unused hardware you have available for squiding.
Re:why not a software solution? (Score:1)
Re:why not a software solution? (Score:2)
I thought about this and have run two copys at once before. If you have any kind of traffic though, you will need a pretty beefy machine with a good chunk of RAM. By the time you have set all this up with hardware costs and time spent, $400 could look pretty cheap.
Re:why not a software solution? (Score:1)
Re:why not a software solution? (Score:2)
I actually have a P3-550 with half a gig of RAM sitting next to my main PC so I am a victim of my own argument... I could probably sell it but it wouldn't go for much more than $300...
Re:why not a software solution? (Score:3, Insightful)
Now I want ot get DSL & Cable, and use both at the same time.
That's not a problem that LVS solves for you.
That's the kind of thing this box does.
This is about the home or small office user making use of multiple internet connections efficiently and easily for their networking needs.
Yeah, of course you can do this with linux... but lvs isn't it.
Re:why not a software solution? (Score:3, Interesting)
http://www.rainfinity.com/products/rainconnect.htm l [rainfinity.com]
Runs on linux, does other rather clever things (can rewrite DNS replies as well for *inbound* load balancing). It works nicely with either a commercial firewall (checkpoint/raptor) or IPTables; or can be used just as an HA router in front of existing firewalls. A feature this crowd will like- you can do everything via a command-line interface if you don't like GUIs too!
DB
Re:why not a software solution? (Score:1)
Re:why not a software solution? (Score:2)
For incoming connections, if both IPs are static you can simply run dns on each IP and roundrobin between them. Will work fine for most purposes.
Re:why not a software solution? (Score:2, Informative)
Think about what would happen: say you started up a ssh session to foo.domain.com; the remote server would see packets coming from two different IP addresses claiming to be alternating for the same session. TCP just doesn't work that way normally.
So as I implied, you'd need to fix your IP inside a subnet that is broadcasted on the BGP routing tables.
Re:why not a software solution? (Score:3, Interesting)
Because not everyone has the time/engery/experience/hardware necessary to set this up on a Linux box. I was running my home firewall on a Linux box until I got one of these things. It has issues, but it generally works and requires less fscking with.
I've had one of these since October, and they're not bad. I got one of these and one of Nexland's wireless access hubs as "review units." I wrote up a review [phoneboy.com] on the product, which details the pros and cons of these devices.
-- PhoneBoy
"I say live it or live with it." -- Firesign Theatre
Load balancing (Score:3, Insightful)
I just can't stop laughing.
Re:Load balancing (Score:2, Interesting)
Why would this make cable/telecom companies "beat their heads" over this? It gives them more business. In fact, I bet it would increase their business. Joe Blow orders *two* cable modems because he wants twice the bandwidth. Same wit DSL.
Yes, for redundancy, you'd be better off with one cable and one dsl, but still, that means that there will be more business for the big guys overall.
Re:Load balancing (Score:1)
Maybe, Maybe not (Score:1)
Of course, maybe this bandwidth fiend would now only be downloading Linux ISO's and lawfully purchased pornographic movies 12 hours a day, instead of 24, and paying more of his/her share.
But I have a feeling these routers will be much more attractive to small business customers. I'm not really sure how pricing schemes for business accounts go. But I once worked for a company with 20 employees, and about 30 computers. They had about 15 computers on one cable modem, and the rest on the other modem. I think a router like this would be attractive to such a office.
YHBT (Score:2)
Re:Load balancing (Score:2)
YHBT YHL HAND
Re:Load balancing (Score:1)
Re:Load balancing (Score:2)
This will not get you faster download speeds. For that you would have to have something arranged with your ISP. What this will do is divide up the computers (or possibly separate TCP sessions on the same computer, I'm not sure) between the two broadband connections. This will let two computers each max out one pipe, instead of having to share the pipe. It won't let one computer use both pipes at the same time.
More LVS info for those interested... (Score:5, Informative)
You can use NAT to hide the real servers from the Internet if you like. This allows you to use most any web server you like (such as IIS), but more fancy routing tricks can be done with Unix or Linux servers for even better results. We use NAT at our site (university EE department) and it can handle more load than we will ever receive -- our objective is high-availability. Also, you can use different methods for different server clusters on the same director (e.g. tunneling tricks for Linux apache servers, and less magic for IIS).
And LVS can be set up such that once a user connects to a particular server, his subsequent connections go back to the same server.
Useful links:
$10 per additional IP addy on Road Runner. (Score:1)
Load Balancing/ Failover (Score:1, Informative)
I've got two BB connections (it's great working for an ISP/Bell) and 1 inside.
The inside connection is secured via NAT and ipchains. The two outside connections are secured via ipchains. I dual-default route out, with some static routes for preferred connections.
Cost me a few hours and a free p-133.
Re:Load Balancing/ Failover (Score:1)
Good lord.... (Score:2, Insightful)
Second, 2 WAN connection AND modem/ISDN backup is sweet for an out of the box solution. Not a bad price, as already stated.
However, and third, a regular PC with a DFE-570tx or it's successor, the 580tx, by Dlink, allows 4 10/100 ports per pci slot. And regular 10/100 nics can be found for less than $10 shipped. You could build a machine for about half the price with greater future expandability.
Re:Hmmmm........ (Score:1)
It should be noted however, that this router cannot load balance a single TCP session across both links, so the maximum you can get for a single TCP session is the speed of a single link. (I think that you possibly could get more outbound speed if the router used source address spoofing, but this may cause problems if the ISP has anti-spoofing filters so they probably don't do this.)
As each link will have a different source IP address packets for a given session will always have to be transferred on the same link.
This also means that if a link fails some sessions will drop. The router will be able to re-establish them over the second link but it will not be seamless as it would with a true BGP connection (but hey, the price is a lot less!)
Inaccuracy (Score:3, Informative)
Network collisions occur when two hosts try to submit simultaneously. The NIC listens for the resulting static on the network line (as static is produced when the signals garble), waits a random length of time, and retransmits. This happens (I believe) at a lower-than-protocol level.
Re:Inaccuracy (Score:2, Informative)
It does wait a random amount of time, but if another collision is detected then the wait time doubles, and the process continues. It's called exponential backoff.
And ethernet protocol (the "physical layer" protocol, in OSI or TCP/IP language) is called Manchester encoding [google.com]. It places 0-to-1 or 1-to-0 transition in every bit, so it's always possible to sync up even in long periods of identical bits.
Re:Inaccuracy (Score:2)
Re:Inaccuracy (Score:2, Insightful)
Collisions can happen on a switch any time two packets (or more) are generated on the wire at the same time. This could be the switch itself and the host at the other end of the cat5. It can happens often on a busy segment (you don't *want* it to happen often, but...).
The original quoted description of collisions is just wrong. The collision light on an Ethernet device has absolutely nothing to do with IP addresses.
A.
Re:Inaccuracy (Score:2)
If each port on the switch connects to a single node on the network and connections are duplex, no collisions will take place. (But imagine if one of your switch ports is connected to a hub with two computers connected to the hub.)
-Kevin
Re:Inaccuracy (Score:1)
Collisions can and do happen on a switch, but not nearly as often as hubs.
Re:Inaccuracy (Score:1)
-Kevin
Re:Inaccuracy (Score:1)
Yeah right (Score:2)
Seriously, if you are a business and have a heavy traffic load or really need a good connection, you don't use broadband... That's good if you have a medium traffic load or need a somewhat reliable connection. I would never trust a broadband connection to be fully reliable, unless it's a dedicated pipe.
And I guess that router is only for normal surfing, no servers. If it constantly switches between two connections, the IP must switch too, right?! I guess one could have a DNS set up with the two IPs but if one of the connections go down, the dns lookup will find the invalid IP every now and then, making the web-site or whatever being run a bit unstable. So this is not a solution if you want to run a service behind it, only several clients using a lot of bandwidth that needs to be load-balanced.
Re:Yeah right (Score:2, Funny)
Re:Yeah right (Score:1)
Ok, so the SLA isn't as good, and 90% of the problems have been LINX routing issues, the other 10% being the fact I'm using DSL that runs over BTs DSL ATM network - Apparently being on an unbundled exchange I can request to be switched over to Easynets own DSLAMs at the exchange, but I've never had an outage lasting more than 30mins.
Ho hum.
OKay (Score:2, Insightful)
This is for, say, having 2 internet connections and using them both. Getting cable & dsl at home and making use of both of them.
Cheap way to upgrade! (Score:2, Interesting)
In the UK at least, the basic home service is 512k down, 256k up and a single IP address. The cost of 1mb down 256k up is much more than twice the basic cost, presumably because it is counted as a business service. Getting 2Mb down 512k up is a lot more again. It would be far cheaper to get 4 lines converted to ADSL with the added bonus of some redundancy.
As far as I know the pricing is set for market segmentation rather than for any inherent extra costs for the fatter pipe. The same home user is unlikely to hog the extra bandwidth, they will just get a better service.
Anyone know any real objections to this from the telcos perspective?
Re:Cheap way to upgrade! (Score:2)
Re:Cheap way to upgrade! (Score:1)
Re:Anything but cheaper... (Score:1)
Looking at the technology suggested, they are talking about linking 2 ADSLs. So if you have 2 lines already in the house like we do it might make some sense to go up to 1Mb down 512k up.
I can't see myself doing this though as the performance of my basic line is fine.
Heh.. (Score:1)
When I first read this, I thought it was an ad directed at all the servers that Slashdot has wiped out. I was about to congratulate Taco for generating a revenue stream. Heh.
Re:Heh.. (Score:1)
Buisines plan without "???":
Read slashdot, wait for sites to get slashdotted
Offer bandwith to slashdotted sites
Profit
Can this be done with Linux? (Score:2)
As for failover, that would be really easy to do regardless of the load balancing support. You just need a cronjob that checks if one of the connections is still up, and reconfigures routing & firewall on timeout.
Re:Can this be done with Linux? (Score:2, Informative)
Here's our linux software solution:
http://www.rainfinity.com/products/rainconnect.htm l [rainfinity.com]
This software uses a linux kernel module that does some neat tricks with packet rewriting to do nat, inspect & modify DNS server replies, nat rules, etc. It also has a configurable connection monitoring service & a bunch of recommended deployments for HA email/web serving/outbound surfing, etc. Works on Solaris and Win2k too..
Re:Can this be done with Linux? (Score:2)
1. IPTables Reference. You will be using the MARK rule, and one of the new modules that do % of time matching.
2. A working knowlage of the IP Route 2 tools.
3. Properly configured Interfaces. You will have one route that will ALLWAYS be primary, then a Secondary Interface. The secondary will will have a slightly higher metric for the default route, but you will need to "src" the packets leaving that interface, and makesure your nat rules are working properly.
4. You need to know your shit to do this. Fucking with this stuff will fuck up your access.
You need no cron job to check the interfaces. Routing does it all for you. Thats why there is this thing called metrics.
-LW - LW@LWolenczak.net
Pathetic Testing Methodology (Score:2, Insightful)
The half-life pings aren't telling us anything, as it's a well-known fact that pings jump when your connection is saturated. It doesn't matter if you're multiplexing two of them.
Win2k/XP can both report raw ethernet throughput using perfmon. This would have been a much more useful and reliable benchmark.
Too many issues are left unaddressed: does this solution double your upload or download rate to a single host? Are you accessible through a single IP, and if so, which one of your broadband connections is used for this?
Can anyone who's actually used this provide some insight?
Re:Pathetic Testing Methodology (Score:1)
In may ways this doesn't do anything i can't do with iproute2 already.
Re:what the f*** (Score:1)
If you are buying a router for $60 to $90, why would you need it to support more than 254 hosts?
Re:what the f*** (Score:1)
Plus:
The device is meant for a home user with maybe a MAX of 10 machines...
if you have anywhere near the 100 hosts range you should be looking at a real router like a Cisco or a properly configured Linux box.
NexLand Security (Score:2, Insightful)
I have complained to Nexland technical support numerous times about the DNS problem. I purchased the router several months ago, and all they've come up with so far is the obligatory "try the new firmware" (which didn't solve the problem).
I would not recommend purchasing the Pro800Turbo at this time, as the hardware/firmware is just not good enough yet (and the tech support is not able to compensate for this shortcoming).
I am now on a multiplexing BSD implmentation (OpenBSD), the two feeds are load balanced pretty nicely (and using just an old P5 box). I don't believe I'd ever go back to the NexLand box again. Also, Linux people might be interested in load balancing in their kernels, I've not tried it msyelf, if someone has please let me know if it's worth looking into.
Mark II from Net Integration Tech (Score:1)
I also evaluating a box that seems to clearly kick butt on the Pro800Turbo. This box is called the "Mark II" from Net Integration Technologies out of Canada. I have one here in my lab right now and testing begins this afternoon. If you want to see it, go to:
http://www.gdbsolutions.com/netitech/markii.
So far I haven't seen any satisfactory black box solutions for under a grand. If you want to do it right, spend a little more money and be pleasantly surprised.
Re:NexLand Security (Score:1)
Re:NexLand Security (Score:2)
It doesn't handle DNS packets very well at all. I have a Debian box set up here with BIND. Unless I point the forwarders at the inside IP of the Nexland, DNS won't work. Why? Because it appears to re-write all the DNS packets from the Internet with it's own internal IP. Nexland's response? That's the way it's supposed to work. sigh
-- PhoneBoy
"I say live it or live with it." -- Firesign Theatre
What we have here... (Score:1)
What broadband users need is something like MPPPoE(Multilink PPP over Ethernet). This is something that I proposed, several years ago, while working for one of the top three network vendors. Marketing determined that there was insufficient demand and it was never implemented in any of the equipment ISP or CPE. There are a few obscure vendors that claim their products do this. But, in order for this to work the ISP must support it at their end. To date, I am not aware of any ISPs that do.
My Netopia does the same deal. (Score:2)
Spammer (Score:1)
Because Software isn't Free? (Score:2, Insightful)
Software isn't free. It requires hardware. When you get dedicated hardware and software that can be configured by someone who doesn't frequent slashdot, you've got a compelling solution.
Anyway, I installed this box at a client site four months ago (two Covad DSL lines), and it's been flawless the entire time. I highly recommend it for situations where better bandwidth isn't available. It's about as easy to configure as a Sonicwall, not quite as easy as a Linksys. Web managed with a gotcha or two in the UI.
I don't understand how this works (Score:2, Informative)
As far as I know, to even do that with big connections you need to go through the same ISP and PPP bond them together. Say I have two T1 lines, one from Sprint and one from UUNet. Each one can transfer 1.54 megabits per second, theoretically. Even though I have two T1 lines, if I go and connect to some remote FTP server, it's only going to send data back to Sprint or UUNet. It can't figure out "hey this guy's got two connections, I should start sending him data on both of them" and suddenly be able to download twice as fast, can I? I may have two T1 lines, but I still can't transfer a file faster than 1.54mb/s.
If if you have two T1 lines from the same ISP (say I have two from Sprint), it takes special configuration, putting them together with a PPP bond, to make them work as one pipe. As far as I know.
Now apply this logic to the type of connections you might have in your apartment. Say you have one DSL connection and one cable connection. Are they really going to increase your transfer speed?
I can see how you'd be able to SEND data faster, but how does receiving work? Can someone explain this to me?
Re:I don't understand how this works (Score:3, Informative)
To do what you are referring to would require a professional router (Cisco, Juniper, linux box with fancy software, etc) with BGP support and ISP(s) that are willing to help you. To use more than one ISP will require your own IP block assigned from the ARIN (not from your ISP's own block of addresses). Work with both of your ISPs to configure routing tables and away you go. Sprintlink, Worldcom, AT&T, Cable&Wireless are very helpful in configuring such a multihomed setup. Below is a link to some info from Sprintlink:
http://www.sprint.net/faq/bgp.html
Re:I don't understand how this works (Score:2)
Symantec Firewall/VPN Applience (Score:1)
The 100 model runs for $365.84 [cdw.com] but could probly find it cheaper than that.
Why Not a business Model solution (Score:2, Interesting)
So how about the bandwidth doubling idea? Great, but wouldn't it be better if the ISPs just changed their business model on cable modems? They already have with DSL. With DSL you could just upgrade to a higher level of service (more bandwidth) instead of consolidating two lower bandwidth lines? With Cable modems, the situation is even simpler. At the modem level, the bandwidth is almost always throttled back. Doesn't it seem idiotic to consolidate two bandwidth throttled lines instead of just opening things up a little? How bout 3Mbps instead of 1.5 (for most AT&T subscribers).
It just seems inane to come up with a hardware or software solution for something that's really a business model issue.
This is timely for me as... (Score:1)
So far I've just used the DSL by setting up a few static routes. Load balancing would be great, but I'm not sure I want to pay $400 for a black box. Correction -- I'm sure I do not want to pay $400 for a black box. I have an ancient P5 serving as my dedicated NAT/firewall and it's probably time to update the kernel to 2.4, stick a fourth ethernet card in there, and dive into the complexities of 2.4 iptables. I would also like to set up some prioritization so that, for instance, my SSH sessions don't stall and my Vonage [slashdot.org] VOIP service doesn't get all choppy when I've got a couple of heavy downloads running at the same time.
I know 2.4 kernel is capable of all this and I've found a fair amount of documentation already, but I wonder if anyone here has any suggestions or pointers to a streamlined configuration procedure or free software package to do this?
Kiscica
Re:This is timely for me as... (Score:1)
http://lartc.org/ you can find this really simple way of doing the samething.
Personally i didn't really care about getting twice the bandwidth only sometimes. So when i did my sharing of one cable and one dsl between 7 housemates i decided just to route people over each link depending on their usage habit. All the web browers people went on my line and all the downloaders got the slower dsl. And of course i bandwidth shaped so people had soft limits and I had no limits. So when i want bandwidth it's out of the way people! All in all it worked quite well.
Re:This is timely for me as... (Score:2)
This seems dandy. (Score:1)
Doesnt make sense (Score:1)
Symantec Version.. (Score:1)
What is this dogshit doing on Slashdot? (Score:1)
Did anyone read the article before posting it? Jeez, I could've gotten the same exact info from reading the side of the box and then reading the instruction manual.
You call that a review??? (Score:2)
All through the several pages there are dozens of spelling, grammar, and simply sentences that just don't make sense. That's not to say that I don't mind that, but in this case the content was the same -- a bunch of screen shots with related commentary of someone who on a good day can setup a Linksys router with no issues. I especially like his "(router talk)" parentheses explanation as if to explain the mystic Mbit unit of measure to us simpletons.
Please, leave the detailed screen shots for the manual and the self-serving explanations to someone with can do more than double click on his Dell. We want to see why we should buy this thing in the first place and how it performs, not how to configure it.
As I skimmed over the first several pages looking for graphs I was instead greeted with some very scientific tests of ping time from within a multiplayer game. Then the guy goes on to download two random files from a random location on the Internet as a testament toward the performance of the router, using a web browser.
No technical or scientific consideration was found in this review, and I found it insulting to read. If you must review something, at least know a little about what you're reviewing, and especially how to test it. Don't waste your time reading that nonsense. In fact, I am surprised it was posted to slashdot considering the quality and the background of the reviewer.
This guy should go back to reviewing the newest shoot 'em up or writing up the procedures for overclocking his celeron, and stay away from stuff that is ever so slightly more complicated involving more sophisticated testing and technical reporting.
Best-way routing (Score:2)
Make sure your ISP isn't putting your traffic through a cacheing server, or this won't work.
Quad Bandwith (Score:1)
I have one of these...my impressions (Score:1)
Software Solution (Score:1)
I was about to buy one (Score:2)
Explanation (Score:2, Insightful)
For example: once the two lines are plugged in, when you are surfing around and hit a website, the router automagically picks which connection (DSL/cable) the request goes out on. If one of the connections happen to be down, it picks the one which is not (thus the load balancer part).
One interesting thing to note though. It may actually seem like the sum of the two connections from an application standpoint. Examples being web browsing and ftp'ing.
In web browsing, I know that in IE and Mozilla, you can select the number of outgoing connections that the browser will use in fulfilling a web request. So you could end up getting the http reponse (text-only) from one connection and using the other (seperate outgoing web request) to retrieve the images on that page. In most cases, you would likely speed up graphics heavy pages quite a bit.
In ftp'ing, some of the clients (along with the download managers) allow you to use multiple tcp streams to receive your downloading file. The software has a file to receive which it starts multiple receives going. In theory, you could run say, 1 tcp stream per connection, and be receiving the same file over the two connections independently, but achieving an overall rate equal to the sum of the two speeds.
The whole thing kinda reminds me of the pigeon-hole principle in a wierd sorta way.
But anyway, I imagine a linux/BSD solution to be cheaper (given low-end hardware requirements).
E
I have one of these (Score:1)
DO NOT BUY THE NEXLAND PRO800T (Score:2, Interesting)
Warning: we have heavily tested the Nexland Pro800T. The Nexland Pro800 Turbo +hard+ crashes daily and looses packets. Once a week it looses all its config. We have had the box replaced multiple times - no help. We have tried their old and newest firmware. No help. It is getting so bad, that Nexland actually shut down their user forums (see www.nexland.com) because so many people are complaining!
I +do+ not recommend the Pro800 Turbo router. The only way we can keep the thing up is to have an automatic ping/tcp/http tester that power cycles the darn thing when it crashes multiple time per day.
Anyone else experiencing these issues?
There is another option. Compex has redundant + load balancing router (NP15-BR). See:
http://www.cpx.com/proddetail_b.asp?c=Broa...%2
Anyone use this?
Hope this helps,
Marc
Re:DO NOT BUY THE NEXLAND PRO800T (Score:1)
Re:Great for office use. (Score:1)
Re:Great for office use. (Score:1)