Choose a Link Balancer

ISP Outage

Anybody with any IT experience will tell you that at some point, you will have ISP issues. Depending on the issue you could be down anywhere from just minutes to hours. I once had an issue with a typically reliable provider where I was experiencing about 25% packet loss. While they were always quick at resolving up or down events, resolving something in between like my packet loss issue took them over 12 hours.

ISP Redundancy Using a Link Balancer

So what does a small company do to mitigate ISP failure? Being a small company does not mean that you cannot have redundancy, but once you sign up for a second ISP, how do you have your network switch over automatically when there is an issue with one of your providers? While there are many products that will give you outbound link redundancy, if you intend to host any important systems in your office, you will want to consider inbound link redundancy.  In the old days, to handle inbound link redundancy, one was forced to purchase two larger routers and hire somebody to configure them with BGP.  But following my premise that keeping it simple means better uptime, I would suggest looking into a link balancer.

Increase Bandwidth with Multiple Carriers

Not only do you gain redundancy by using multiple carriers, but you get to aggregate the bandwidth for all of your users to use.  This is significant if you have limited options in your service area.  If only DSL is available in your area, by bonding multiple DSL lines together, you can aggregate up to some serious bandwidth.

How Does a Link Balancer Know that an ISP is down?

Ecessa-Testpoints

The Ecessa products allow you to setup specific test points for each ISP WAN link.

Using the Ecessa Clarilink 250 as an example of how most Link Balancers work, each WAN ISP connection is setup with defined testpoints that are continually accessed to verify that they are reachable.  Options such as ICMP vs TCP connection options are available as EcessaTestRatioswell as a definition of what timeouts are acceptable.  Furthermore, an acceptability ratio of responses to #tests allows you to define how many dropped packets are acceptable in these tests.  This helps one to overcome situations like the packet loss issue I mentioned earlier.  While I used the Ecessa Clarilink in this example, there are many vendors in the link balancing industry.

Choose a Link Balancer

There are several vendors in the (inbound) link balancer space.  Several options in the industry are:

  • Barracuda Link Balancer has three different models to choose from depending on the throughput you are needing.
  • Ecessa has three different product lines:  Powerlink, ShieldLink (adds firewall and VPN to Powerlink feature set), and ClariLink (Adds SIP Call Balancing and Failover to the ShieldLink feature set).
  • Elfiq has products from the LB-600B upt to the mammoth LBX5500.  Some of their products include their Inbound Geolink features which are pretty cool.
  • Fatpipe markets its WARP and XTREME products as making your WAN 300% more redundant reliable and fast.
  • Peplink’s webpage advertises their Model 20 – 1350.  Inbound link balancing begins with their Model 210, and Firewall and the ability to manage their Pepwave wireless access points begins with the Model 305.

Each of these products have similar basic features including Link Balancing and failover, Traffic aggregation, SNMP, VPN, Inbound DNS balancing, and QOS.  While some of their products add firewall, hardware failover, SIP Balancing, and geographical WAN balancing to their product.

 

Seeing it in Action
The Diagram below shows a basic setup with a typical link balancer.  The company is a small business that hosts both a webserver and an ftp server on premises.  They have contracted with two different ISPs to handle their internet traffic.  All of their users can use outbound traffic:  browse the web, send email, chat or facebook without knowing which ISP they are being routed through.  The link balancer aggregates the bandwidth from both connections.  When one of the connections fails, all of the outbound traffic is routed through the other ISP automatically.

To handle the inbound traffic:  people visiting the company website, or inbound ftp connections or perhaps receiving email messages from customers to their on-prem email server, they had to setup inbound failover.

Setting up Inbound failover

The diagram below shows that both NAT rules and a DNS Zone file are setup on the link balancer:

EcessaDiagram

When both ISPs are up, inbound traffic say to the web server (www) will query the DNS services on the link balancer and get a response of one of the two WAN addresses (x.x.x.4 or y.y.y.4) setup in the NAT rules.  The web client will then use that address to connect to the web server.  The TTL (time to live) on the DNS response is set very low so that if a failure occurred, the client would quickly query the DNS again and get the other address for the web server.

Other Things to Consider

I was a bit skeptical of how well it work when I purchased my Ecessa.  But the results have been fantastic.  Simpler than BGP routing, my job has been made easier and my uptime has been better because I purchased a link balancer.  The biggest things that you get from a link balancer is control of over your on premises hosting by providing simple inbound ISP redundancy, redundant outbound internet or VPN connections to other offices or partners.  The only real issue I have with link balancing is that it does not provide is multi-site redundancy for on premises hosted systems except when using the highest end of some of the product options.  But it does offer an affordable solution to small business to resolve their ISP connectivity nightmares and that is what uptime by simplicity is all about.

The following two tabs change content below.
Jeff has 20 years of professional IT experience, having done nearly everything in his roles of IT consultant, Systems Integrator, Systems Engineer, CNOC Engineer, Systems Administrator, Network Systems Administrator, and IT Director. If there is one thing he knows for sure, it is that there is always a simple answer to every IT problem and that downtime begins with complexity. Seasoned IT professional by day, Jeff hopes to help other IT professionals by blogging about his experiences at night on his blog: http://uptimemadeeasy.com. You can find Jeff on or LinkedIn at: LinkedIn or Twitter at: Twitter

Latest posts by Jeff Staten (see all)

Leave a Reply

Your email address will not be published. Required fields are marked *