Wednesday, June 30, 2010

Will Telx Produce Better Margins than Equinix?

by David Gross

Collocation provider Telx (TELX), which filed an S-1 earlier this year but has yet to issue shares to the public, is often compared to industry leader Equinix (EQIX).   However, there is a major difference financially and operationally in that Telx leases every one of its facilities, except for Atlanta, while Equinix owns or has capital leases on most of its locations.   Of Telx's 15 facilities,

Is the InfiniBand Bandwagon Actually Growing?

by David Gross

Looks like UK IT magazine The Register is now drinking the InfiniBand Kool-Aid.   They're excited about the InfiniBand Trade Association (IBTA) roadmap to 312 Gbps, and how much faster this will be than the recently ratified 100 Gigabit Ethernet.

One of the points I've been making to people looking at the costs of these technologies is that the defining economic trait

Tuesday, June 29, 2010

Data Center Providers Hit Particularly Hard in Selloff

NASDAQ was down 3.85% today, but data center providers did even worse.   Some of the big decliners today included:


    One company bucking today's trend was CDN provider Limelight Networks (LLNW), which finished up 3 cents.

    Google Network Architects Endorse InfiniBand

    I've made the point in a few recent posts that InfiniBand is sticking around, and hopes that it will fade away like Token Ring, FDDI, or ATM are very much misplaced.    Now in addition to the financial traders and supercomputing centers who've shown strong support for the technology, Google has released a white paper on how to

    Mellanox Announces Record Message Rate for Supercomputing Applications

    Through a test network built with its InfiniBand switches and adapters, Mellanox (MLNX) announced that it has transferred a record 90 million MPI messages per second.   MPI is a protocol used in the supercomputing industry to manage the transmission of data between computing nodes.    While I generally don't like to promote vendor marketing messages coming out of test labs, this is the sort of technical capability, along with low cost hardware, that has helped InfiniBand hold its niche in HPC and high-end data center networks against Ethernet.

    Fabrinet Holding Above IPO Price

    Fabrinet (FN), a Thai contract manufacturer of optical components used in telecom networks and data centers, went public on the NYSE Friday, pricing at $10 a share.  The company's customers include JDSU and Infinera.  In two days of trading it has held above its offering level, although today will be interesting to see how it trades with the weak economic news out of China.

    QXX8WU4D4JJW

    Data Centers Continue to Grow in a Weak Economy

    by David Gross

    Few industries have withstood the recession as well as data centers. Lease rates are still rising, capacity is being filled up, and new centers are opening every week. The industry's great recession was not in 2008 or 2009, but in 2002, when it was decimated by dot com overbuilding, and some facilities were turned back into warehouses.

    Riding the industry's recessionary expansion, Digital Realty (DLR) grew to $191 million in revenue last quarter, a 28% year-over-year increase. The data center REIT is now larger than many of its retail and office counterparts, and is over half the size of office stalwart Boston Properties (BXP). During its first quarter, co-location provider Equinix (EQIX) grew at a comparable 24% year-over-year rate, coming very close to a $1 billion annual run rate before closing its merger with former rival Switch and Data.

    The double digit growth for an industry that suffered during the last recession raises the question of whether this expansion can continue, or if the current economy is too weak to sustain further top line growth. Nonetheless, there are a number of factors supporting the industry right now, including:

    Dependence on financial trading volumes, not underlying asset values. Pushing trades, quotes, and market data across data center networks, the financial services sector accounts for about a quarter of the industry's revenue. And traffic volumes in the recession have had a weak correlation to underlying security prices. As long as the number of algorithmic trades, derivative price quotes, and electronic executions continue to rise, traders are likely to demand more data center space.

    Growth in consumer transactions. Amazon.com (AMZN), an Equinix customer, is still growing its top line over 20% per year. In addition to e-commerce companies, content providers like Hulu and Sony, and many gaming networks, are all running their traffic over data center networks. An indicator to watch here is page views for these consumer sites, as well as the trend of top 10 sites developing their own centers, as Facebook and Google (GOOG) are doing right now.

    Major price discrepancies in long-reach and short-reach data links. Not quite as obvious as some of the consumer indicators are the persistently high price differences between short-reach and long-reach data connections. For example, a 10 kilometer, 10 gigabit link can require more than 10 times the capital outlay of a 10 meter link due to large cost differences in the optical modules and line cards needed to provision the connections. Therefore, it makes far more sense for content providers to interconnect with one another, or with their telecom carrier at a neutral data center, rather than construct a dedicated private line to reach the nearest telecom central office.

    In addition to these factors, data centers have not been impacted much by the lack of job growth. Online video page views don't sag because of a weak unemployment report, and the cloud computing and Software-as-a-Service (SaaS) providers locating servers in data centers are automating many of their customers' job functions themselves. Leading SaaS provider salesforce.com (CRM), another Equinix customer, is growing at a 20% year-over-year pace comparable to Amazon.

    The credit crisis, high unemployment, and limp economy of the last two years have done little to slow the growth of public data centers. If anything, limited credit has stopped the industry from overbuilding like it did eight years ago.

    Monday, June 28, 2010

    Tokyo Institute of Technology Deploying Voltaire InfiniBand Switches

    InfiniBand continues to hold firm in its supercomputing and financial trading niches, with Tokyo Institute of Technology rolling out Voltaire's (VOLT) 40 Gigabit switches in its 1,400 node TSUBAME 2.0 supercomputer. The deployment includes over 4,000 edge switches and 18 director switches.

    With prices as low as $400 per 40G port, these short-reach InfiniBand platforms continue to serve a niche in low latency apps where many predicted Ethernet would take over. While 40/100 Gigabit Ethernet was ratified last week, the traditional data center market is still a few years away from mass 40 or 100 gigabit deployments with either protocol.

    disclosure:no positions

    Number of Publicly-Traded Companies Dropping, but Telx and CoreSite on the Way

    In the data center industry, we've got two major companies in registration, but for the rest of the market, there are a dwindling number of investment opportunities. In a WSJ story today, Bill Hambrecht notes that the number of companies filing proxies with the SEC has dropped from 9,100 ten years ago, to 6,450 today. And the number of companies tracked by the Dow Jones U.S. Total Market Index is down to 4,136, a drop of nearly 20% over the last five years.

    He cites limited interest from i-banks in underwriting smaller IPOs now, and I'd add that Sarbanes-Oxley hasn't helped either. But with Telx and CoreSite on the way, the data center industry is likely to stand out even more among public market investors.

    Will 100 Gigabit Create New Data Center Networking Startups? (Part 2)

    In the earlier post on 100 Gigabit network, I discussed how the shift from designing in ASICs to FPGAs, while seemingly technical, is having major business implications for data center networkers. In this post, I look at the impact on FPGA developers.

    As ASICs become more challenging economically, designers are increasingly turning to FPGAs to handle highspeed packet processing, which is an important development for Xilinx (XLNX) as the leading developer of FPGAs. By trading out the ASIC design cost, and replacing it with a royalty fee for an FPGA, circuit board and system designers are taking out a large capital expenditure and replacing it with an ongoing operating expense, which is the typical outsourcing economic equation. The increase in outsourcing is helping not just networkers, but their chip suppliers maintain some staying power, an important consideration for data centers managers concerned about suppliers' financial viability. Xilinx, for example, has been reporting strong gross margins, close to 65%, and the company has bounced around the 60s for most of the last decade. While this is lower than some other chip makers, including Mellanox (MLNX), Xilinx has made up for it by keeping R&D expenses to just 20% of revenue, compared to 30% of revenue at Mellanox. As a result, it has been able to post consistent net margins of around 20%. Rival Altera (ALTR) has posted similar margins, although its revenue remains about 20% lower than Xilinx’s. One factor behind Xilinx's lower gross margins is that inventory turns for the FPGA vendors are typically 5, which is not much higher than optical components suppliers. While other chip makers simply order wafers that ship to the fab suppliers with whom they have contracted, Xilinx and Altera must stock additional inventory for their distributors because their products require additional programmability before final use.

    With FPGAs, Designers to Replace an Increasing Fixed Cost with an Ongoing Variable Cost
    As ASICs become more challenging economically, designers are increasingly turning to FPGAs to handle highspeed packet processing, which is an important development for Xilinx (XLNX) as the leading developer of FPGAs. By trading out the ASIC design cost, and replacing it with a royalty fee for an FPGA, circuit board and system designers are taking out a large capital expenditure and replacing it with an ongoing operating expense, which is the typical outsourcing economic equation. The increase in outsourcing is helping not just networkers, but their chip suppliers maintain some staying power, an important consideration for data centers managers concerned about suppliers' financial viability. Xilinx, for example, has been reporting strong gross margins, close to 65%, and the company has bounced around the 60s for most of the last decade. While this is lower than some other chip makers, including Mellanox (MLNX), Xilinx has made up for it by keeping R&D expenses to just 20% of revenue, compared to 30% of revenue at Mellanox. As a result, it has been able to post consistent net margins of around 20%. Rival Altera (ALTR) has posted similar margins, although its revenue remains about 20% lower than Xilinx’s. One factor behind Xilinx's lower gross margins is that inventory turns for the FPGA vendors are typically 5, which is not much higher than optical components suppliers. While other chip makers simply order wafers that ship to the fab suppliers with whom they have contracted, Xilinx and Altera must stock additional inventory for their distributors because their products require additional programmability before final use.

    Sunday, June 27, 2010

    Will 100 Gigabit Create New Data Center Networking Startups?

    When MPLS (Multi-Protocol Label Switching) was released in the late 90s, Juniper (JNPR) entered the market with a proprietary packet forwarding ASIC, i.e. Application-Specific Integrated Circuit. When Gigabit Ethernet arrived around the same time, Extreme (EXTR) and Foundry (BRCD) introduced their proprietary packet forwarding ASICs. As 10 Gigabit Ethernet was being ratified in 2002, Force10 hit the scene with its proprietary packet forwarding ASIC. So who will develop a proprietary packet forwarding ASIC for 100 Gigabit? No one.

    As process geometries in the semiconductor industry get smaller, the fixed design costs for each chip increase. Ten years ago, when 130 nanometer was still a state of the art technology, it cost around $10 million in R&D to develop an ASIC for forwarding packets. Today, at 40 nanometers, it costs well over $100 million.

    Regardless of what happens with the economy, VCs fondness for the industry, or the development of the data center market, it is no longer financially possible for a startup equipment vendor to build a carrier-class forwarding architecture with its own ASIC, in addition to having to write an NMS, and sell a box. This is why startups like BLADE and Arista are focused on boxes that have LESS functionality than competing platforms from large vendors, but can also shoot frames across a network with less latency, as long as there aren't too many different places that traffic can go. In some respects, they're like the Model Ts of data networking, simple and inexpensive platforms with fewer configuration options but very strong price/performance.

    In a follow up post, I'll look at what this means for programmable logic vendors like Xilinx (XLNX) and Altera (ALTR).

    Digital Realty Trust - Now Over Half the Size of Office REIT Boston Properties

    Digital Realty (DLR) announced its 2Q earnings call will be held Friday, July 23rd at 1pm EDT. During its 1st quarter call, the company raised annual Funds from Operations guidance 8 cents, from 3.26 to 3.38 a share. Average rents on leases signed during the 1st quarter were $141.

    DLR's 1st quarter revenue of $191 million was a 28% year-over-year increase over Q1 2009, and put it at over half of the level of leading office REIT Boston Properties (BXP), which was flat over the same period. So in a few years, the largest public REIT could very well be a data center provider.

    Saturday, June 26, 2010

    BLADE Network Technologies Announces sub-700 Nanosecond Latency for IP Multicast

    Privately held BLADE Network Technologies announced new, low latency capabilities for its Rackswitch 10 Gigabit Ethernet platform at last week's SIFMA Financial Services Technology Expo. Primarily targeting price quotes and market data for the company's financial trading customers, the enhanced capabilities, as well as product focus, give you some idea why BLADE is hanging around when many pundits expected that larger vendors, especially one beginning with a C, would put it out of business.

    One factor working in BLADE's favor right now is all the R&D invested in Carrier Ethernet forwarding capabilities, and massive memory capacity, have left a few vendors with overcooked boxes that go well beyond the needs of many data centers, which don't need to manage 300,000 route BGP4 tables like public Internet providers. This has also kept a lot of cost out of these data center switches, which are narrowly designed (vendors won't say that, but I mean it as a compliment) around latency needs, and NOT trying to handle every data networking protocol ever created.

    ADC: We're Ready for 40/100 Gigabit

    ADC (ADCT) issued a press release stating its support of the recently ratified IEEE 802.3ba standard for 40 and 100 Gigabit networks. In addition to fiber and copper cables, the vendor supplies optical distribution frames to data centers.

    40 Gigabit InfiniBand is already selling, but many of those links are for short-reach copper interfaces. On the Ethernet side, many data centers are still upgrading from Gigabit to 10 Gigabit, so the standardization is more of a technical milestone than a major business development.

    Facebook's First Company-Owned Data Center to Open in Six Months

    Facebook's Prineville, Oregon data center is set to open in six months. The social media company's primary existing centers are the DuPont Fabros (DFT) facility in the DC area and the Digital Realty Trust (DLR) center in Santa Clara.

    Located in central Oregon, Prineville is about two hours from the new Google facility in The Dalles. High traffic websites have sought out the Pacific Northwest because of its low cost hydro and windpower. Meanwhile, the wholesalers and co-location providers continue to focus instead on access to high bandwidth Internet exchange points in DC/Northern Virginia and Silicon Valley.

    Router Vendors Need to Control Costs to Remain in the Data Center

    The price gap between router ports and switch ports continues to grow, with 10GBASE-LR ports going now for about $4,000, compared to over $200,000 for OC-192 POS ports. And while short-reach products like 10GBASE- SR and 10GBASE-CX are anywhere from 30% to 70% cheaper than 10GBASE-LR, there is only about a 15% savings when putting a 1310nm transmitter on an OC-192 POS card in lieu of 1550 nm transmitter.


    The impact of persistently high router port prices is that everyone outside the major telcos is looking to bypass them, either by pushing more traffic forwarding down to the optical layer, or by stretching point-to-point Ethernet networks to reduce the number of routing hops. While there is an operational benefit of containing IS-IS or OSPF tables, the cost of filling up a network with OC-192 or OC-768 router cards is simply prohibitive for many enterprise and research applications.


    Router cards are not likely to get much cheaper, because many of their costs are tied to hardcoded features that can be done in software at lower line rates. The advent of 10 Gigabit networks has brought along heavy demand for TCAMs and network processors that speed up route lookups, but adding more electronics has only exacerbated the price gap relative to static connections. Add in a few load balancing features or the ability to forward on TCP port number and electronics costs really take off relative to standard Ethernet switches.

    While Cisco (CSCO) has both switching and routing products it can put out there, Juniper (JNPR) arrived late to the top-of-rack switching market, and many of its efforts to beef up memory and operating system capabilities might impress telecom providers, but serve little purpose in the data center.

    Welcome to Data Center Stocks

    In the 2000s, the telecom industry focused heavily on broadband and wireless network construction. In the 2010s, well over half of all households and businesses now have broadband or wireless service. As a result, the networking industry's growth this decade will move back indoors - to the data center, which is already creating new demands for higher speed connections, advanced networking architectures, and products that meet the low latency needs of large enterprises and government agencies, not the long distance requirements of telecom carriers.

    At the same time that network equipment suppliers are growing in the data center, data center REITs are adding new space while traditional office REITs face a slow and mature market. As a result, the data center is one of the few technology sectors that offers opportunities for both the growth and dividend investor.

    In this blog, I'll look at the data center REITs, network equipment manufacturers, co-lo providers, and everyone in the data center supply chain. The emphasis will be on microeconomic analysis more than chasing down next quarter's EPS. While written for investors, I'm also providing product and company depth that should be useful for the industry as well.