Foundry forges ahead of the competition – Vendor
Having the audacity to challenge a hugely dominant supplier on its home turf requires–apart from courage verging on recklessness–the ability to stay always a step ahead, either on functionality or price. In the enterprise backbone market Cisco reigns supreme and, while it may be facing threats from big names coming in from the voice market, such as Nortel, it has managed to defend its marker share from direct data networking rivals while expanding into the newer worlds of wireless and voice over IP.
But there are healthy alternatives to Cisco. Some of these sell their wares on the basis of better pricing, notably 3Com and Dell, coming up from the departmental level, and more recently, China’s Huawei. Others opt for the more difficult route of trying differentiate on performance and technology, picking off niches where cutting edge capabilities are essential. Extreme has gone some way down this road, and more aggressively, so has Foundry Networks.
Founded in 1996, Foundry is one of the few direct challengers to Cisco left. It specializes in Layer2/3 Lan switches with the FastIron and EdgeIron ranges, Layer 3 backbone switches (BigIron), Layer 4-7 web switches (ServerIron) and metro routers (NetIron). It has also recently made its first move into wireless switches. Compared to Cisco, it is tiny, though it has been profitable on an annual basis since 1998 and has a healthy cash pile. Revenues for fiscal 2002 were $300.7m, which was a drop on 2001’s 311.2m. However, the company had been targeting profits ahead of growth, and increased net income sevenfold to $22.5m. In the last fiscal quarter, Q2 of 2003, revenues climbed by 28% year-on-year to $95.7m and quadrupled net income to $16.8m.
Foundry currently gets about 85% of its revenue from the enterprise–with government, financial service and education the biggest verticals–but the service provider market is starting to grow again, particularly in the Asia Pacific region. However, founder and CEO Bobby Johnson admits that both sectors have been tough since 2001, particularly the ISPs. This year, the US–which accounts for 60% of Foundry’s sales–and Japan have started to pick up, he says, and the UK is stable, but Europe as a whole remains depressed.
Given its David and Goliath situation, Foundry needs to focus on uniqueness of technology on both the hardware side–this year, it has introduced its fourth generation Asic chipsets, the Terabit architecture, designed for the migration to 10Gbit Ethernet backbones–and software, where it claims leading edge network management facilities in its IronView Network Manager package and support for the sFlow standard for network traffic monitoring.
Perceived technology leadership is vital to its quest to convince companies to veer from the safe option of choosing Cisco. Market research from the Dell’Oro Group gives Foundry market leadership in three areas: Layer 3 10Gbit Ethernet switches, Layer 3 modular Gigabit Ethernet switching, and Layer 4 to 7 Web switching. This is in a market for Ethernet switches that the same research company estimates will increase in value from $11.3bn last year to $16bn in 2007. The areas where Foundry has chiefly focused on grasping the lead are seen as the highest value subsectors. “Our aim is to get first to market, to be an innovator and to give better price/performance,” said Johnson.
Such a strong focus on leading edge markets is high risk however, requiring heavy R&D investments for a company of this size. R&D spend is between 10% and 12% of revenue depending where Foundry is on the design cycle, especially where Asics are concerned. Johnson says engineers are motivated by working on products that are at the cutting edge of the market and by feeling directly accountable for–and rewarded for–their success. On this basis, engineers are paid lower than average salaries but have generous stock options.
“We have to achieve advanced technology but manage on a relatively small R&D budget compared to Cisco,” says Johnson. “We only hire experienced people and they work long hours, but the experience levels mean we have faster decision making, less trial and error and therefore faster time to market than Cisco.”
This particular corporate culture he has built is important to Johnson and he sees it as a key differentiator against Cisco, just like the products themselves. Balance is key–Foundry demands long hours and high levels of dedication without offering huge levels of perks, but it also rewards successful developments, not just with shares but also “time off for rejuvenation after a project”. “We hire people who want an adventure not a job, and who love technology,” Johnson says with feeling. “We appeal to people who find larger companies too slow and frustrating. We’re more like the Wild West, run and gun, our review process is hours, not weeks or months.”
Foundry will always come up against conservatism and the fact that it is seen as easier and safer for companies to stay with Cisco. Although there is a class of organization that is tired of the dominance of the giant and is pursuing an ‘anything but Cisco’ policy, it is far more common to take the safe route. “For every account out there that would like to move away from Cisco there are many that don’t have the time or intestinal fortitude to do it,” concedes Johnson. “But we do see a significant percentage that wants a second vendor–even if inertia prevents them moving completely.”
THE MOVE TO 10GBIT
Foundry has chosen to differentiate itself from competitors by providing a clearer and more explicit roadmap for companies and operators moving to 10Gbit. This has won it customers–Cardiff University, for instance, says it chose Foundry for its next generation backbone, which will support advanced applications such as simulation, because it was the only “prepared to have an open book on the future and commit to implement new technologies as they become available” (see Rethink IT October issue for case study).
However, the danger of committing to next generation technologies at an early stage, and aiming to be ahead of the curve in rolling them out, is that vendors leap too early and invest in products that nobody wants to buy for another few years. “There is always a danger of coming to market too early but it is one of the things that most attracts people to Foundry. We were a year ahead of the Gigabit and 10Gbit Ethernet standards but promised to replace the kit if it was not compatible. There is only one place for early adopters to go and by the time a technology is mass market we’re on our second or third generation,” said Johnson.
10Gbit is an example of a technology where this balancing act is particularly difficult. The standard is significantly more complex than previous Ethernets, partly because it is designed to support wide area networks as well as enterprise backbones, and so has two physical layers defined–one for 10Gbit Lan interfaces, and one designed to operate at a compatible data rate with the carriers’ Sonet and ATM (9.58Gbps).
Unlike Gigabit Ethernet backbones, which met a huge pent-up demand for bandwidth, 10Gbit has been driven primarily by the vendors and there is currently little user requirement for it. It is starting to be installed to connect high end routing switches, but many companies will make do with aggregating Gigabit links via 802.3ad to scale up their networks, until 10Gbit pricing comes down considerably, which is likely to be mid-2004.
Until late 2004, the main need for 10Gigabit will be among the metropolitan area carriers, which do need to aggregate bandwidth while keeping the cost advantage of Ethernet over Sonet and ATM. Other trends that will slowly drive enterprise applications and uptake are data center consolidation and the increase of 100Mbps or Gbit links to the desktop. Some storage area network vendors are also looking at 10Gbit as a replacement for Fibre Channel, although this is unlikely to become widespread for another two years.
One example of the kind of contract that Foundry seeks out as playing to its strengths is a recent one with one of the US’ largest educational district authorities, Clark County, Nevada School District (CCSD). The organization has contracted for a high speed metropolitan area network to carry voice, video and data to its 289 schools, enabling a range of new teaching methods including distance learning and real time videoconferencing for ‘e-schooling’.
More than 100 schools are now linked to the new network and full roll-out will be completed within a year. The network is already carrying 21m web requests and 100Gbytes of network traffic each day. By 2012 it is expected to be serving 405,000 students and their teachers and families.
“We are getting well ahead of the technology curve for voice, video and data education instruction–historically, school districts are used to playing catch up,” said Dr. Philip Brody, chief technology officer for CCSD. “This network is already beginning to deliver profound benefits to the educators, students and administrators of CCSD as we provide teachers with the data, graphics and video necessary to advance student learning. Two important human benefits of the technology are improved communications between schools and parents by placing phones in every classroom that connect over an IP network, and improving school-to-school resource sharing. CCSD expects to recoup much of its capital costs on this network investment through significant operational savings in coming years.”
The network is based on 600 servers and 300 Lans connected by Foundry’s Gigabit Ethernet backbone, based on BigIron and FastIron switches. CCSD has upgraded from a Frame Relay architecture and says that extending Ethernet end-to-end from the Lan across the backbone ensures high performance, increased efficiency and low latency.
Clark County also uses Foundry’s IronView Network Manager.
MOVES INTO WIRELESS
The only area where Foundry has reversed its usual approach of leaping first into the market is wireless, where it has been quite late to announce switches. It finally entered the fray in September, taking a similar technical approach to Cisco’s.
The company claims its cautious timetable has enabled it to wait for standards such as 802.11g for fast Wi-Fi and the interim security specification WPA to be well established, so that customers do not have to buy kit and then upgrade it again very quickly. Like Cisco, its key selling point will be the integration of wireless with its wired backbone networks and unified management of the two, something the WLan specialists clearly cannot offer. This leads it naturally to favor the fat access point approach, where most of the intelligence is incorporated into the AP, rather than the architecture of wireless specialists like Trapeze, which provides almost dumb APs with the intelligence residing in a central switch.
TOO BIG A COMMITMENT
Phillip Kwan, director of enterprise applications at Foundry, says the centralized switch approach requires too big a commitment at a time when WLan is still new technology and IT budgets are still low. “Most companies prefer to implement on a small scale first and grow gradually,” he said. “A Trapeze switch is $12,000 but all the 24 ports are dedicated to WLan, so you have to use them all to get value. You can’t run voice over IP, videoconferencing, VLan tagging and so on in these switches as you can with a Cisco or Foundry switch.”
Kwan claims coming late to the market means Foundry is offering state of the art technology, although much of the differentiation argument against Cisco focuses, unusually for Foundry, on price–it is pricing its APs at $400-$500 less than the Cisco Aironet 1200 and is using more powerful processors, running at 260MHz compared to Aironet’s 200MHz. The radio chips come from Atheros. One of the key advantages that Foundry claims is support for ‘Clear to Send Only’ technology, which reduces the way in which dual-mode 802.11b/g networks have their performance dragged down by ‘b’ clients.
The new products, grouped under the IronPoint label, consist of the IronPoint 200 802.11a/b/g access point, the WiFi Management application suite; and a WLan software upgrade option for Foundry’s FastIron range of switches. This combination enables users to implement small standalone WLans or to manage them from an existing FastIron wired infrastructure.
For the first time, Foundry will target small and medium enterprises as well as large companies, claiming that its standalone APs will offer SMEs a more robust alternative to traditional SoHo products such as D-Link’s, with support for quality of service and other features, for an affordable price premium. The AP is priced at $795 and the management software at $1,195. This will be the primary route, initially at least, to using the WLan products to expand Foundry’s customer base, rather than just providing a new option to its existing customers.
Among those customers, Foundry says 35-40% have said they would implement a WLan of the type the supplier is promising within 12 months, which gives the company a target of 2,000 to 3,000 users in the first year. 80% of its base are enterprises and 20% are ISPs and metro service providers. Johnson stresses that the main aim of the wireless range is to help penetrate further into some enterprises rather than to be a standalone business, and he seems less interested than Kwan in moving down-market to the SMEs, aware that there it will meet 3Com and Dell and have to adopt a whole new pricing structure that would sit badly with Foundry’s high value, high margin technology.
This does not mean that Foundry will not see the SME specialists knocking on its enterprise door. There is a real threat to Cisco, Foundry and Extreme from these suppliers, particularly Dell, with its strong enterprise contacts and legendarily efficient manufacturing and supply chain. And now Chinas Huawei, having settled legal difficulties with Cisco, is also entering the US market.
Although these lower cost vendors tend to focus on less sophisticated areas of the networking range than Foundry’s key products, it cannot ignore them entirely, especially in the price-conscious enterprises. Forrester Research believes that Cisco and Foundry still beat Huawei and the other low cost options on features, on offering integrated solutions rather than point products, and on providing a low risk, trusted choice for CIOs.
However, pressure from these suppliers will force the established names to adapt their strategies. Forrester, for instance, believes that Cisco and Foundry should cease trying to differentiate themselves on the basis of chip prowess, especially with commodity silicon readily available from third parties, and should instead look at feature innovation, clear return on investment and business value, and strong reliability and quality of service.
On the more traditional front, the main vendors apart from Cisco that Foundry comes up against are Extreme, mainly in Germany but rarely in the US; Nortel “occasionally and mainly in France”; occasionally Riverstone or Alcatel. But most of these suppliers have retreated back to their installed bases as far as switches are concerned.
Names like Alcatel and Nortel will be far more visible to Foundry as voice and data converge in the enterprise. “Convergence is the next big driver,” says Johnson, but as usual Foundry is looking for a slightly different angle on the trend and one that depends on advanced technology. While voice over IP will be an essential strand of any networking supplier’s strategy, Cisco already has huge share in this developing market. Foundry wants to focus on the video aspect of convergence–“integration of voice and data will go mainstream, but there is also video and in certain applications that will become more and more important,” says Johnson. “Video will be more important than voice for some applications.”
The focus on video is logical given that it will drive the installation of bigger backbone pipes and so drive Foundry’s core business. It is also enabling the company to build up a strong reputation and market share in particular niches where video over IP is already important, notably video and film production and the academic research world.
It illustrates the strategy that Foundry must adopt going forward if it is not to be crushed by the huge weight of Cisco. It must continue to identify sectors and individual enterprises that require maximum network capacity for specific applications and are therefore prepared to take a risk on a smaller vendor, need to buy equipment before it becomes mainstream–at a premium–and choose technological advance over the safe route. Convergence and the revival of the service providers will both help make this strategy work, but as enterprise budgets remain tight, Foundry needs to ensure that its products are not overkill even for its advanced users, and that it branches into new areas such as wireless in order to keep its revenue options somewhat open.
COPYRIGHT 2003 Rethink Research Associates
COPYRIGHT 2004 Gale Group