Fail over with AutoFailover

Mark Gibbs' Web site tips, plus network applications news headlines

Load balancing and failover are crucial features in any Web service system, and you can spend a lot of money creating such systems. Today, we have an alternative that is definitely worth looking at if you are trying to keep online service costs low and keep the amount of work you have to do as small as possible.

The service, from Tzolkin (see editorial links below), is called AutoFailover. I first became aware of Tzolkin some years ago when I was researching Dynamic DNS (DDNS). Tzolkin's DDNS product, TZO, is one of the leaders in this field with Linksys providing built-in support for TZO in its Etherfast DSL/Cable routers.

Anyway, AutoFailover is a simple idea: You specify two to four servers on different Internet backbones and AutoFailover becomes the authoritative DNS for them. After that AutoFailover continuously monitors the availability of whatever services you are concerned about from two different points on the 'Net. Should both monitors report that a server is down, all requests are directed to the next available server by dynamically modifying the DNS entries.

To continue reading, register here and become an Insider. You'll get free access to premium content from CIO, Computerworld, CSO, InfoWorld, and Network World. See more Insider content or sign in.

Load balancing and failover are crucial features in any Web service system, and you can spend a lot of money creating such systems. Today, we have an alternative that is definitely worth looking at if you are trying to keep online service costs low and keep the amount of work you have to do as small as possible.

The service, from Tzolkin (see editorial links below), is called AutoFailover. I first became aware of Tzolkin some years ago when I was researching Dynamic DNS (DDNS). Tzolkin's DDNS product, TZO, is one of the leaders in this field with Linksys providing built-in support for TZO in its Etherfast DSL/Cable routers.

Anyway, AutoFailover is a simple idea: You specify two to four servers on different Internet backbones and AutoFailover becomes the authoritative DNS for them. After that AutoFailover continuously monitors the availability of whatever services you are concerned about from two different points on the 'Net. Should both monitors report that a server is down, all requests are directed to the next available server by dynamically modifying the DNS entries.

AutoFailover can also respond to DNS name resolution requests and provides each of the servers a load balancing service in a round-robin style. You can also "weight" your servers so those with a higher load-handling capacity can be preferentially allocated for requests.

When one of your servers in a load-balancing configuration goes down it is removed from the rotation system and added back in when it's back up.

This load-balancing system can also work with whatever load balancing you might be using to front server farms so multiple farms on different backbones are load balanced and supported by failover.

AutoFailover can monitor all the standard protocols (HTTP, HTTPS, SMTP, FTP, telnet, PING and POP3) and other protocols by request. If AutoFailover detects a protocol failure, you can have failover for just that protocol or for the entire domain.

Tzolkin claims that the maximum time required to redirect server requests is 2.5 minutes, which includes failure detection, DNS record changes, and DNS propagation time through other DNS servers with an average time of 1.25 minutes.

AutoFailover pricing starts at $99.50 per month or $995 per year for two servers. There's a list of options to enhance and extend the services.

Read more about software in Network World's Software section.

Mark Gibbs is a consultant, author, journalist, columnist and blogger.