How to deal with discontinued gear

Think about your options before equipment is discontinued.

James Wiedel faces a predicament: Cisco has announced it will stop selling and supporting its ONS 15540 dense wavelength division multiplexing platform, and he has three of them in his network at the University of Southern California.

It’s not a pressing problem, because the last day of sale for the gear is Jan. 31, 2006, and it will be supported until Jan. 31, 2011, but Wiedel has already started thinking about his options.

And that is just the right thing to do to carry out a smooth and effective transition from old gear to new, industry experts say.

When a vendor decides to pull the plug on a product, it can leave business customers scrambling for a replacement. But if customers are well prepared, the demise of the gear can be an opportunity to upgrade to less costly, more efficient products that sport more features.

Wiedel, the director of networking at USC, has already looked at newer DWDM gear that costs tens of thousands of dollars as opposed to the ONS 15540s, which cost more than $700,000, he says. Ultimately, he plans to keep the equipment in the network until he finds it lacks the features that USC needs.

Being well prepared for product end-of-life is key, says Jean Ritala, secretary of the IT Service Management Forum USA. The group proposes that IT departments should adhere to strict processes when it comes to the life cycle of hardware and software, with most hardware scheduled to be replaced every three to five years. That way when notice comes that a piece of equipment is being dropped, chances are it will already be scheduled for replacement before the vendor suspends support for it, she says.

When an end-of-life situation does arise, IT executives and those who hold the purse strings are then in a better position to deal with it. “It’s important so there’s no surprises, especially devices that cost a lot of money. [Financial] executives don’t want surprises,” Ritala says. Preparation lets IT build upgrades into future budgets rather than trying to squeeze the funds out of current budgets by canceling other purchases, she says.

Part of the process is setting up a database of what gear is in the network and how it is configured, and possibly using software tools to automate the task, she says. If automation is used, it is also essential to regularly check a random sample of devices by hand to make sure the locations and configurations match what’s in the database.

Good data of this type enables network executives to act quickly when they are hit with an end-of-life notice for a particular device, says Trent Waterhouse, vice president of marketing for Computer Associates . “How many do I have and where are they? If I see I have 100 of them and 90 of them are on the latest version of firmware, let me make sure I get the last 10 up to the last-known supported version,” he says.

An updated inventory has helped out Wiedel, who faces a more pressing challenge than replacing his optical equipment. Entarasys has ended support for about 100 SmartSwitch 2200 switches that were in his network, and he’s in the midst of replacing them. Enterasys also is dropping the Ethernet blades he has in more than 800 SmartSwitch 6000 chassis across the network. But he knows where they all are and how many he has, which has made planning the upgrade easier, he says.

So far he has pulled most of 2200 switches from his network and moved the rest to less demanding spots. He already began moving to Enterasys’ next-generation Ethernet blades, so he scoured his network for unused ports on those blades and consolidated them as much as possible, freeing up several blades altogether. So far the cost has been minimal and he has not budgeted for a wholesale replacement.

When the discontinuation notice arrives, IT executives have to decide whether to stay with the same vendor or seek another, Waterhouse says, which should include a look at how well the previous equipment performed. “Look at the records, and if you see more trouble tickets associated, consider a new vendor,” he says.

IT departments should always research options from other vendors, says Zeus Kerravala, vice president of enterprise networking at The Yankee Group. “If you’re going to retire a product, do a comparative cost evaluation. It gives you price leverage if nothing else when you negotiate with your current vendor,” he says.

In Wiedel’s case, Enterasys had performed well and had an upgrade path that matched Wiedel’s plans. It is dropping its old hardware cards for the 6000 series chassis but will support new ones. Fitting the old chassis with the new Ethernet blades rather than buying all new switches was attractive. “It saves a whole bunch of money,” he says. “We just do a blade swap, and you may only be changing out one card in a chassis,” he says. He would not detail the costs.

The upgrade presented an opportunity to solve problems that had been looming, Wiedel says. For instance, USC was upgrading its wireless authentication, which required virtual LAN features only found on newer Enterasys Ethernet cards, he says. So switches linked to the wireless gear have gotten the upgrades, but others can wait. “We’re started on a slow migration,” he says.

A gradual transition is a good approach because it gives IT executives the chance to spread out the cost decide whether it makes sense to shift around the old equipment to less important locations, Kerravala says. “So if it’s a data center switch, you might move it to the wiring closet. If it’s a core switch, move it to a big branch” he says.

Vendors usually give notice well before they stop selling gear, so if a business decides it needs to keep the product in the network past the support date, they have the chance to buy more while they’re still available and keep them around for spare parts, Kerravala says.

Whatever replacement decision is made, USC’s Wiedel says, the most important thing is to make sure the transition keeps services running and disrupts end users as little as possible. One strategy for that is making small-scale swaps to learn how to avoid pitfalls. “Take a test net and try to do a conversion on that and find all the little bugaboos you’re going to get. You’ll never know what isn’t going to work and what is going to work until you try,” he says. “Plan big, start small.”