It’s a WAN optimization controller, Jim. But not as we know it: Minnesota-based Ecessa’s PowerLink WAN optimization controller (WOC) range is unlike other offerings in the WOC market because it doesn’t carry out compression, data caching and application acceleration, or packet shaping, data monitoring and bandwidth management.
In contrast to market leaders such as Blue Coat Systems or Riverbed, Ecessa takes a completely different approach by suggesting, in effect, that you can indeed solve many WAN performance problems by throwing bandwidth at them. “The problem with accelerators is that you end up getting latency, and when you add compression, you just exacerbate that,” says Jason Breyer, Ecessa’s VP Sales. Blue Coat, Riverbed and others in the WOC market would probably dispute these claims.
Anyway, Ecessa’s idea is that small and medium sized organizations – those with anything from a dozen or so users up to about 1,000—can cut their WAN costs, increase WAN reliability through redundancy, and increase their overall WAN bandwidth, using a PowerLink device. Essentially the PowerLink aggregates multiple WAN links—which can be any combination of up to 15 connections including DSL, T1/E1, cable, frame relay, wireless and satellite—so they look like a single WAN link. It then carries out load balancing: When traffic is generated from the corporate network the device looks at the available links and routes the traffic through an uncongested one. If an individual link gets too congested or goes down, the device stops routing traffic to it and redirects it to an alternative link.
This has a number of benefits. Perhaps most importantly, it brings a degree of redundancy to an organization’s WAN link. Instead of using a single T1, a company could use multiple DSL connections, or a T1 and a DSL line. If one link goes down, the WAN would continue to operate, albeit with reduced bandwidth, using the other links. “SMBs can get rid of costs by going with broadband connections with lower SLAs instead of getting another T1,” Breyer points out.
To ensure that traffic can still come in to an organization’s servers when a link goes down, the PowerLink device also acts as the primary and secondary authoritative DNS name server for an organization’s domains. If a link goes down the device stops providing that link’s IP address when receiving DNS queries, and instead provides the IP address of an alternative link. Since the host name record is given a TTL of just a few seconds (usually 30), the IP address of a failed link is only cached on other servers for a very short period before being refreshed with a working IP address.
When the failed link is restored, the device will advertise this to caching servers which will update their records the next time they make a query after their cached information expires.
This technique can also be used for site failover for business continuity and disaster recovery. Essentially, a PowerLink device at a secondary location queries a similar device at the primary site at regular short intervals. If the primary site fails to respond, the secondary site assumes control of the corporate domains by updating DNS records so that all traffic is redirected to this site.
While the Ecessa devices don’t carry out compression, caching or acceleration, they do carry out a rudimentary degree of traffic management using quality of service (QoS) for outbound traffic. “If you can identify a packet by protocol, packet length, port or source or destination IP address, you can assign it a priority level or minimum bandwidth,” says Breyer. “But really we don’t do much management. It’s better to have multiple WAN links.”
The economic case for Ecessa’s PowerLink devices (or its new ShieldLink range of devices which also include a firewall and VPN gateway) appears to be strong in the appropriate circumstances. They range in price from $1,295 for the PowerLink 50 which supports two WAN links and a single inbound service, to $5,995 for the 250EHQ which supports up to six remote sites, up to 15 WAN links and up to 512 incoming services. A support contract is about $1350 for three years. “The ROI is rarely over a year,” says Breyer. “The savings come from using one or more broadband connections at $70 per month, instead of a T1 at $500 per month, or scaling from a T1 to a T1 plus one or more broadband connections instead of getting a second T1.”
The PowerLink devices lack the sort of sophisticated packet shaping capabilities to prevent all an organization’s available bandwidth becoming clogged by rogue peer-to-peer or other unwanted traffic, and they do nothing to speed up applications running across WANs from branch offices to servers in a central server room. To that extent the functionality that the devices offer is not what would normally fall into the class of WOCs. But if they function as they are intended, they certainly provide a way to add network redundancy and site failover to an organization at a very reasonable cost.