Alaxala's stated goal is to develop backbone routers and switches, but the company's offerings extend down to two- and four-port switches (more on that later). The company got its start in June with ¥5.5 billion (about $50 million) from its parent companies. Hitachi owns 60 percent of Alaxala, with NEC holding the other 40 percent. (See Hitachi, NEC Create Core Router JV and Hitachi, NEC Form Router Company.)
Obviously, Cisco and Juniper stand in the way of Alaxala and anyone else trying to crack the router/switch market. That's why Hitachi and NEC decided to combine forces, creating Alaxala (which has no known connection to fellow Japanese palindrome OOIOO). Based in Tokyo, the company employs 300 and is headed by CEO Hiroyuki Wada, who comes from the Hitachi side of the family.
The idea was to produce more reliable routers and switches, a cry that soon-to-be Alaxala officials were hearing from customers, says Kazushige Arai, Alaxala's general manager of business planning.
"Service providers and large enterprises want to build more reliable networks that can be used all the time, as a lifeline service," Arai says. "So, Hitachi and NEC decided to build a new platform to cover such requirements."
Alaxala also provides an Asian-based alternative to North American vendors, says consultant Dan Golding of the Burton Group. "The buzz in the carrier community is that [Alaxala] didn't want to build a more reliable router; they wanted to build a more 'Japanese' router," he says. "I don't think they're in a strong technical position, but if they can get anything going, they're in a good cultural and social position."
Alaxala will start by selling in Japan, where it has the home-field advantage against not only Cisco and Juniper, but also Chinese upstarts Huawei Technologies Co. Ltd. and Beijing Harbour Networks Co. Ltd. From there, the company hopes to branch out to other geographies.
Alaxala has three product lines due to ship by the end of January:
- The AX7800R line of backbone routers, with four-, eight-, or 16-slot models available. The largest of the routers, the AX7816R, has a backplane switching capacity of 384 Gbit/s and can handle up to 32 ports of 10-Gbit/s traffic if XFP modules are used (it's 16 ports for larger modules). Prices range from $207,000 to $234,000, based on a conversion rate of ¥111 to $1 U.S.
- The AX7800S multilayer switches, which also target the backbone. Again, the line includes four-, eight-, or 16-slot options, with a maximum of 384 Gbit/s of switching capacity. In terms of interfaces, one notable difference from the 7800R is that these switches don't have modules with XFP interfaces yet. Prices here range from $87,000 to $103,000.
- The AX5400S multilayer switches. These are low-end boxes, in two- and four-slot configurations, with prices between $35,000 and $40,000. Switching capacity for the 5402S and 5404S is listed at "about" 24 Gbit/s and "about" 48 Gbit/s, respectively. Both systems can handle Gigabit Ethernet interfaces but not 10-Gbit/s Ethernet.
On the router side, the high-end AX7816R would appear to be set against core routers such as the Cisco 12816 or the Juniper T640.
The obvious question is whether Alaxala can truly beat the reliability of established routers and switches. NEC and Hitachi have produced switches and routers before, but Golding notes that their forte has been at the commoditized end of the market. Alaxala is promising features such as graceful restart, and the AX7800R router series is capable of non-stop software upgrades – that is, new software can be loaded without having to reboot the machine. More advanced steps, such as a modular OS, remain on the drawing board; Arai says the company is "examining the potential of the technology."
Alaxala intends to draw on NEC and Hitachi engineers to develop its systems, but it would theoretically be faster to just buy the technology. And that brings in the inevitable Procket Networks Inc. angle. "There was speculation when Procket was in play that Cisco purchased it to keep it out of the hands of NEC and Hitachi," Golding says. "The Procket technology wasn't necessarily superior to anything Cisco and Juniper do, but it was a legitimate third way." (Core-router startup Procket is still in the process of being acquired by Cisco – see Cisco to Pay $89M for Procket Assets.)
A Procket pickup would have given Alaxala a nice head-start, Golding thinks. "As Procket found out, carrier requirements are very extensive. You need a lot of functionality, and it takes a long time to write the code," he says. He thinks Alaxala is "going to be challenged" if it can't acquire an extra dose of technology from somewhere.
— Craig Matsumoto, Senior Editor, Light Reading
For further education, visit the archives of related Light Reading Webinars: