Catch up on stories from the past week (and beyond) at the Slashdot story archive

 



Forgot your password?
typodupeerror
×
Hardware

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.
This discussion has been archived. No new comments can be posted.

Nexland Pro800Turbo Load Balancing Router Review

Comments Filter:
  • by User 956 ( 568564 ) on Sunday June 30, 2002 @04:03PM (#3796633) Homepage
    Why not a software solution, instead of dropping 400 bucks? Ultra Monkey is a package including LVS, prepared mostly by Horms.

    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.
  • by User 956 ( 568564 ) on Sunday June 30, 2002 @04:08PM (#3796651) Homepage
    For those interested in using LVS for software routing, it's fairly simple. Basically, you patch a stock Linux kernel and use a tool similar to ipchains to establish virtual services. These services forward requests to your back-end real servers according to a flexible ruleset that you design.

    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:

  • by Anonymous Coward on Sunday June 30, 2002 @04:11PM (#3796661)
    I've been doing mostly this (manually though) via my linux-based fw/router.

    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.
  • by march ( 215947 ) on Sunday June 30, 2002 @04:20PM (#3796684) Homepage
    This will not load balance "a site" - rather, it will load balance your connection to the internet.

    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.
  • Inaccuracy (Score:3, Informative)

    by acrhemeied ( 316269 ) on Sunday June 30, 2002 @04:21PM (#3796688) Journal
    "If the Duplex LED is flashing this means their has been a collision on your network. This happens when packets are dropped for some reason or the packets have been misdirected. This usually only happens when two computers are using the same IP address and this usually only happens when you specify an IP address rather than using the DHCP feature built in the router."

    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.
  • by JPriest ( 547211 ) on Sunday June 30, 2002 @04:35PM (#3796732) Homepage
    Why would you need the BGP info? Both broadband connections are routing to the exact same gateway router. You are just load balancing the data over 2 lines to get there.
  • by Hoonis ( 20223 ) on Sunday June 30, 2002 @04:51PM (#3796780) Homepage
    Yes, there is a feature in ip route 2 that lets you set up load balancing out multiple connections. It's a bit of a muddle to set up & make it work with IPTables but worth the effort. Setting cronjob to test if things are "up" is a little harder than it sounds though, a dying T1 line often is up 30% of the time, which is actually reason to leave it working for inbound email etc..

    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:Inaccuracy (Score:2, Informative)

    by ddstreet ( 49825 ) <ddstreet@ieeOPENBSDe.org minus bsd> on Sunday June 30, 2002 @05:23PM (#3796876) Homepage
    ...waits a random length of time, and retransmits. This happens (I believe) at a lower-than-protocol level.

    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.

  • by sean@thingsihate.org ( 121677 ) on Sunday June 30, 2002 @05:48PM (#3796966) Homepage
    Can someone explain how this works to me?

    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?
  • by bozoman42 ( 564217 ) on Sunday June 30, 2002 @06:12PM (#3797063) Homepage
    This actually wouldn't work, because you don't have an autonomous system number for a subnet (and AS's aren't given below about /23 anyways). In other words, you'd need real routing protocols like BGP on your border router.

    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.
  • by green pizza ( 159161 ) on Sunday June 30, 2002 @06:31PM (#3797120) Homepage
    From what I understand, this $400 gizmo is geared towards homes and businesses that mostly surf and download on multiple computers. Unless some black magic is used, no *single* download or upload will exceed the thruput provided by the faster of the two internet connections. It's simply not possible... the router has been assigned two IP addresses from two totally different ISPs that don't know anything about each other. It is similar to a single threaded application running on a dual CPU computer -- the application will only take advantage of one CPU. However, if there are more threads running, they'll be balanced across the two. Same goes for this router. If you have multiple downloads running, they'll be spread across the two internet connections at the router. It may not be the end-all solution, but it sure would be handy for a download-happy household.

    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

A morsel of genuine history is a thing so rare as to be always valuable. -- Thomas Jefferson

Working...