The All IP Enterprise

IP hegemony didn't happen by accident. Consolidating as many services as possible into an IP-based network makes good business sense. It simplifies your architecture and eliminates the headaches that come from managing disparate equipment and protocols. The more systems you move to IP, the more efficiently you take advantage of staff expertise.

Take VoIP (voice over IP). Thirty-eight percent of readers responding to our poll for this package say they have an enterprise VoIP story in production beyond pilot stage, and 33 percent say they plan to implement one within 24 months. What's the attraction? As we discuss in "IP PBXs Come Into Their Own", rather than paying a telco engineer to perform adds, moves and changes, your helpdesk can assume these functions.

Michel Labelle, manager of network and terminal support services for Terminal Systems, a Canadian port management company, says that a previous employer saved about $150 (Canadian) per change by shifting this function to the helpdesk. And, the time to complete the change dropped from days to minutes thanks to the ease of VoIP provisioning. Other components of the VoIP infrastructure, such as servers, also can be delegated to the network staff.

The biggest obstacle to VoIP? In our poll, initial upgrade cost led the list, followed by the need for network improvements, which we discuss in "Prioritizing VoIP Traffic: Can You Hear Me Now?". Moving to an all-IP network may cut costs, but not necessarily from shedding the staff that managed the TDM infrastructure. "If you get rid of a telecom engineer, you probably have to bring in a senior network engineer," Labelle says. "The savings are a wash."

Instead, consider folding staff into the network group and investing in training. Even if they are being moved to a field outside their core expertise, they still have a deeper understanding of your network and business than a fresh hire. "A telco engineer can transition into the network group," Labelle says. "They understand the network topology, cabling and infrastructure. You're talking $10,000 of retraining to get someone productive again."

You can also take advantage of their existing knowledge base. "Even if you move a telco engineer to the network team, you still have areas like cabling and voicemail that match up to their skill sets," he adds.

The staff issue is slightly different when it comes to storage. Although companies are willing to swap out an analog telephone system for VoIP, as we discuss in "iSCSI Takes on Fibre Channel", no enterprise will dump its investment in FC in favor of iSCSI just to move to an IP-based storage model. What's more likely is that companies evaluating network storage for the first time have an incentive to choose iSCSI over Fibre Channel.

Why? First, there's the cost and complexity of deploying and maintaining a FC SAN. Ethernet costs about $200 per 1-Gbps switch, compared with $1,500 per port for a 2-Gbps FC switch port. And, FC complicates the network architecture because it requires a separate infrastructure of switches, HBAs (host bus adapters) and management tools, all of which must be run by staff with specialized knowledge of FC operations--and often, hefty salaries to match their skill sets.

By choosing iSCSI you can use Ethernet switches and draw on your staff's experience with IP networking. And thanks to 10-Gbps Ethernet, iSCSI now compares favorably with FC on the performance front; for more on 10 Gig and its role in the all-IP enterprise, see "10GigE: Fast, Pricey and Coming to a Data Center Near You".

Not surprisingly, 10GigE is poised for growth: 442,000 ports will be sold by the end of 2006, according to The Dell'Oro Group. That's up from just 6,800 in the first quarter of 2004. Prices are also forecast to drop substantially (see chart).

However, iSCSI does have drawbacks. The cost of 10GigE switches, around $3,900 per port, negates some of the price difference between FC and Ethernet products. 10GigE also runs best over fiber, which means expenses beyond per-port costs. In addition, Ethernet adds overhead that can affect performance. Specialized NICs and HBAs can improve throughput and reduce the drain on server processors, but there may be a cost penalty.

For companies that haven't yet invested in an FC SAN, iSCSI is a compelling option that deserves investigation. We recently reviewed modular iSCSI SANs and found that these systems are an affordable way for small and midsize businesses to consolidate multiple NAS or DAS systems. And while enterprises should consider them for departmental and remote offices. Clearly, those contemplating upgrades to an existing FC SAN now have a credible alternative, which at the very least can be used as leverage when negotiating with FC vendors.

At present, FC has the edge when it comes to high-speed storage needs, because 10GigE is simply more expensive. But prices for 10GigE are half what they were two years ago and will continue to fall. And that means the all-IP enterprise has time on its side.

Andrew Conry-Murray is Network Computing's business editor. Write to him at acmurray@nwc.com.