One of the more interesting infrastructure themes currently getting attention and investment is Software Defined Networks. The ambition is for there to be a simple network control plane, with clearly defined APIs, enabling changes in network operations depending on the priorities of the applications currently using the network.
The post collects my notes.
Updated 3 May 2012 to add summary from Google keynote at the Open Networking Summit in April.
IBM white paper (May 2011) describes IBM and NEC Openflow demo at Interop 2011.
"The OpenFlow Protocol allows entries in the Flow Table to be deﬁned by a server external to the switch, which creates the potential to unify server and network device management. For example, a ﬂow could be a TCP connection, all the packets from a particular MAC or IP address, or all packets with the same VLAN tag. Each ﬂow table entry has a speciﬁc action associated with a particular ﬂow, such as forwarding the ﬂow to a given switch port (at line rate), encapsulating and forwarding the ﬂow to a controller for processing, or dropping a ﬂow’s packets (for example, to help prevent denial of service attacks)."
Nanog 54 - the most interesting panel at San Diego (6 Feb 2012) discussed Openflow, standards work in progress, and one approach to implementing SDN.
Quoting Ed Crabbe (Google) on why Google is interested in SDN - it sees an opportunity to improve cost control for their infrastructure while maintaining performance.
common threads: partition resources and control within network elements
minimize network element local control plane
offline control of forwarding state
offline control of network element resource allocation
minimize complexity of software local to network elements
SDN has been around as a concept for a long time. Ipsilon GSMP, 1996; Cambridge's The Tempest, 1998 and so on down the years. Flowspec in 2008 looks like openflow, too.
IETF PCE, 2004, based on having RSVP-TE deployed, ISIS, etc. From network element software side, it's more complex, but it's deployable today.
But why SDN?
comes down to cost. Most of what we do is cost optimization with bounds for performance
make efficient use of resources
network element CPU and memories
underlying network capacity
move heaviest workloads off expensive, relatively slow embedded hardware to fast, cheap, commodity hardware
reduce network complexity and thus operational overhead and outage time
simplify policy composition
enforce correctness constraints and invariants
reduce inter-dependencies between routers, between protocols
reduce complexity of distributed control system software on network elements
implement innovative new techniques (Heller's Elastic Trees, etc.)
Innovation velocity - desirable to speed feature implementation and deployment
Do new protocol development fast; right now, takes too long to get new protocols into devices.
IBM distributed virtual switch
IBM announced a distributed virtual switch (5000V) 14 Feb 2012 supporting SPAN, ERSPAN, Ethernet Virtual bridge, load balancing, ACLS, etc, integrated into VMware. Alex Bennet (Battery Ventures) "This is an important development because as the first tier of network switching moves into the server, the virtual switch becomes extremely strategic real-estate and control point for emerging SDN architectures."
Summary and notes from Urs Hoelzle's keynote at ONS, titled 'Openflow @ Google'
"OpenFlow has helped us improve backbone performance and reduce backbone complexity and cost"
For the WAN, aim is to have cost per bit/second go down with scale, as for CPU and storage; without a lot of systems engineering, that's not what actually happens, since network complexity goes up; also, a 100G bps interface costs a lot more than 10 x 10 gbs interfaces, or 100 x 1G bps.
Google are operating the WAN backbone carrying traffic between data centers with centralized traffic engineering.
Multiple switch chassis in each domain, custom hardware running Linux; Quagga BGP stack, ISIS/IBGP for internal connectivity
They have a simulation environment for testing "with the complete system, end to end. Everything is real except the hardware".
Building a software defined WAN results in higher performance, better fault tolerance, and lower costs.
Conclusion : Openflow and SDN are ready for real world use.
Last Wednesday, 1 Feb 2012, was an all day mentoring session for the finalists in IBM's global Smartcamp competition. It was held at the old Federal Reserve Bank building in the financial district in San Francisco. Mentors, of which I was one, spent 40 minutes with each team, having first listened to their pitches. IBM provided a summary of each company. My notes on each are in bold. If you are interested in more background, contact me at email@example.com .
Finalist Company Information: from IBM
1) BitCarrier www.bitcarrier.com
Bitcarrier is the leading provider of real-time traffic information for smart cities. Its patented Bluetooth and Wi-Fi scanning technology provides very accurate and reliable traffic information in real time.
By collecting locations from an average of 48% of target vehicles in real time, Bitcarrier has created the biggest database yet of citizens’ travel behavior. This information is the basis for decision-making regarding new infrastructure or mobility policies, such as the provisioning of public transportation, but it also has unlimited potential from private companies in services/applications they could provide to clients and cities.
Stored and managed in Bitcarrier’s private cloud, the data collected becomes valuable information about congestion generation, the origin and destination of travelers and commuting times.
Bitcarrier’s integrated platform is the seed for new deployments of traffic sensors in the city. It provides an integrated gateway for other technologies like outdoor parking and loops and it provides an open data stream platform and an advanced visualization tool for data dissemination. With Bitcarrier’s integrated platform, city managers can check several KPIs to understand the real benefits of newly implemented policies, and private companies have the infrastructure to develop services to solve end-users’
everyday problems. Bitcarrier’s vision is shared and used by cities such as Barcelona, Rio de Janeiro and São Paulo, and by some partners like Telvent-Schneider Electric.
Barcelona, Spain. Selling to highway toll operators - limited market - and to city governments in places which have little existing traffic control infrastructure and do have money. This means mainly Latin America. They aim to become the communications platform for other infrastructure status information.
2) C-B4 www.c-b4.info/jsite
C-B4 provides unique pattern-based predictive-analytics software that helps organizations make smarter decisions to improve their business results. C-B4 technology is fully automated and does not require experts’ support on
a regular basis. It provides vertical solutions for consumer products and retail customers, as well as a horizontal predictive-analytics server that is easily integrated into telecom, industrial and security systems. C-B4 server weaves its patterns from the actual data stream, and thus provides a unique pattern fingerprint for any object of interest: customer, product, process, service, website, user or more. C-B4 models identify hidden patterns, with practically no limitations on the number of hierarchies and
dimensions. As a result, C-B4 is innovative in three primary ways. It predicts changes much earlier than other solutions enabling actions sooner and reducing costs. It identifies anomalies and opportunities that others miss improving services levels. It identifies complex interrelationships among customers, stores, products, services and other key variables
These activities are performed automatically, with no need for expert intervention, and provide clear, actionable results. The company has leading customers such as Proctor & Gamble, Kimberly-Clark, Applied Materials, Lely, Cellcom, Tnuva, Delek Group, Strauss, General Motors, the European Community, and several intelligence agencies.
Maskit St. Herzlyia, zip code 46733 Israel
Pattern based predictive analytics Academics out of Tel Aviv in 2000
10 person company, good match for Watson
3) ConnectM connectm.com
ConnectMs vision is to be a leading energy management solutions provider leveraging BI and analytics across telecom tower and building infrastructure.
The ConnectM offering is an end-to-end solution that provides remote monitoring and energy management of telecom infrastructure for energy spend optimization, operations and management, asset management and revenue
assurance. ConnectMs Energy management solution for commercial buildings monitors, optimizes and controls energy systems (e.g., lighting, HVAC, UPS, etc.).
The market for solutions that manage and optimize energy, as well as operational spends in the major target verticals, is estimated to be $16B (U.S.) globally. ConnectM leverages its current leadership position in India, with an install base of 5,000 cell sites, while aligning with global players to build a scalable and profitable business. A significant player
in commercial buildings in India, with an installed base of 3 million square feet of office space, ConnectM delivers annualized energy savings in excess of $4M (U.S.) to current customers, including leading Indian firms as well as global MNC companies.
4th Floor, KMJ Arcadia,No.15 (Old No.310/2),5th Block, Koramangala, Bangalore 560095
$6m funding in 2007 (Crunchbase) from IDG Ventures India (media biz, venture fund) and Sasken. These people need expansion capital, not venture - they have a repeatable installation and ROI model.
IDXP analyzes consumer behavior in real time for the brick-and-mortar retail market. We track the paths to purchase and analyze shoppers’ behavior at the point of sale through an innovative technology that uses smart sensors in shopping carts to monitor their movement in real time. We generate online reports, insights and premium information to help
manufacturers and retailers make better decisions. We allow them to strengthen their marketing initiatives, raise their product conversion rates with instant results, and improve the overall shopping experience for consumers.
Raja Gabaglia, 2708 - Sala 133 - Estoril. CEP. 30.350-540. Belo Horizonte, MG - Brazil
Selling paid pilots to retail associations - groups of grocery stores, where one will be the pilot and the others will follow on. Some government funding. Sales process is 4 - 6 months.
Localytics offers the most powerful application analytics platform; giving mobile app publishers for iPhone, iPad, Android, BlackBerry and Windows Phone 7 the tools they need to build more successful and profitable applications. Localytics provides the only real-time service, session-level detail and data access demanded by top application publishers.
One Cambridge Center, 6th Floor, Cambridge, MA 02142 USA
Funded $3.2m, 700k 5/10, $2.5m 4/11
Localytics provides mobile usage data and analytics for the mobile market, similar to companies such as Flurry and Medialets.Customer lifetime value is being optimized. http://www.localytics.com/blog/
Pricing Free; $100, $1000 /month.
8000 apps, 150m users - good video, on Vimeo
These people don't necessarily need any more capital - they are doing very well both in traction and in revenue.
6) Palmap palmap.cn
Palmap is a pioneering indoor mapping and POI open platform. Palmap aims to redefine indoor maps and bring new values to mobile users walking inside buildings rather than those driving on streets. We are seeking a breakthrough for the popular LBS applications, where our smarter, super-detailed maps and POI information will greatly help mobile users with indoor activities such as navigating airports and shopping. Based on Palmap’s innovative indoor GIS engine and open APIs, LBS developers could build numerous creative and interesting LBS and SNS Apps such as check-in, store locator, around-me search, location-based games, etc. Palmap incorporates the cutting-edge technologies of GIS, cloud computing, mobile computing and sensor networks to help make our planet smarter.
405, No.88 Lane 777, W Guangzhong Rd, Zhabei Distric, Shanghai 200072, China
Not locating individuals on maps in real time - providing (very detailed) map information to partners, who may. Gathering information by sending paid groups to malls, airports; also by crowdsourcing for updates. Has raised money from China and Hong Kong. Jacky Zhou (formerly Baidu) just joined as Chairman.
7) Profitero www.profitero.com
Profitero is a next-generation pricing intelligence service for retailers and manufacturers. The company helps clients to increase sales and maximize profits by leveraging high-quality online competitive data at scale. This allows businesses to react quickly to changes in their competitors’ prices. Some of the world’s biggest retailers already use Profitero pricing data.
Retailers and manufacturers have found Profitero’s data essential for pricing strategy, forecasting, ongoing price management, merchandising planning, product promotions and market/brand positioning.
The technology currently monitors 27.5 million products across 2,500 European retail websites with plans to grow this to 100 million products in 2012. Suited to online and brick-and-mortar retailers and manufacturers, Profitero’s easy-to-use web dashboard application can readily compare accurate pricing data across grocery, drinks, health and beauty, household and electrical goods – including A-brands as well as own-label lines.
Kill Ave, Dun Laoghaire, Co.Dublin, IRL
18 Dec 2011 - raised €750k series A, Delta Partners; Enterprise Ireland & Bank of Ireland
Development center in Belarus. These people won the competition.
8) Secure Waters secureaqua.com
SecureWaters, Inc. developed and commercialized patented technology that monitors detects and identifies toxins in surface water. We manufacture and sell the AquaSentinel, an electronic monitor/alarm system for continuous protection of drinking water sources. AquaSentinel is the first system of its kind to automatically and continuously test for chemical toxins in source water supplies using indigenous algae as the biosensor without the need for consumables or reagents. The system allows for early intervention counter-measures against potential threats or accidental contamination for source water assets typically used by water utilities for drinking water. This real-time chemical/biological detection system with remote sampling
and sensing capability significantly improves the safety of the nation’s water supply by ensuring a continuous stream of high-quality water. The sensor can be deployed in multiple units to create a network of interconnected devices, while reducing costs by replacing manual testing, providing 24/7 coverage, eliminating costly cleanup of water utilities and eliminating costly regulatory fines. All of these solutions solve demonstrated pain points our end-users have identified as needing a solution. Our proprietary software and its easy-to-use GUI interface can be integrated into municipal water facility control systems, allowing automatic and timely system protection and cost-savings.
8701 Dayton Pike, Soddy Daisy, TN 37379 USA
Web site needs a couple of case studies, from viewpoint of operators and purchasers. 1m pitch vid shows wireless transmitter, pictures of green algae. Self funded so far, could do with venture investment. Have licensed exclusive patents from Oak Ridge National Labs.
2012 is shaping up to be a mentoring year. Complementing the Cultivate Companies work, I'm being a mentor for Techstars Cloud, which is running in San Antonio until April. The first face to face meeting with the seed companies there is later this week.
At the beginning of February, IBM runs its global final for SmartCamp - nine companies from nine cities, each of which won a regional final. I'm one of the mentors for the global final event in San Francisco.
Austin: SecureWaters – water monitoring solutions
Bangalore: ConnectM - energy efficient telecommunications solutions
Barcelona: BitCarrier – traffic management solutions
London: Profitero – analyzes on-line competitive information for retailers
Istanbul: Skin Scan – pocket scan technology for skin cancer prevention
New York: Localytics – real-time data analysis for mobile applications
Rio: IDXP – smart sensors in shopping carts to monitor consumer behavior in real time
Shanghai: Palmap – Indoor and POI Data open platform
Tel Aviv: C-B4 - Context based analytics for hidden data patterns in large scale data
Blog posting has been sparse, though those of you who follow @annejohn will have seen occasional reports about Scotland from a couple of weeks back. With Tom Willey and Richard Marshall, I was in Edinburgh and Glasgow to meet people and talk about Cultivate Companies. What we are aiming to do is to take advantage of geographically separate demand and supply - in Scotland (and elsewhere) there are smart people developing interesting technology, but funding sources are limited and have a low risk appetite. In Silicon Valley there's a shortage of engineering talent, and many sources of investment.
Cultivate Companies aims to match scare resources with demand.
Here is Edinburgh Castle, with the McEwan Hall, taken from an office with one of the better views in the world.
Techstars is a startup accelerator - like Y Combinator, except that it isn't in the Valley; it runs in Boulder, Boston, New York and Seattle. In January, the first Techstars Cloud will run in San Antonio, Texas.
Entry is competitive; "TechStars will select approximately ten companies focused on cloud infrastructure to participate. Chosen companies will receive seed funding, mentorship from many of the world’s top cloud infrastructure thought leaders, and will have access to perks available only to TechStars companies and alumni."
I'm one of the mentors for the Cloud event; accordingly, I participated remotely in Techstars for a Day, which ran on Saturday 5 November 2011, as a sort of introduction. In addition to running a video feed from the venue, there were several Google+ Hangout sessions, with a machine in a room next to the main hall in the venue (inside the hall would have been much noiser) and up to 10 of those of us 'attending' remotely participating in the hangout. Several of the applicants introduced themselves and talked about what they wanted to do if they were selected.
Some people had camera and microphone issues on their end; the chat session was useful for checking on background and asking who did and didn't have signal. For me, the audio and video quality of the hangout was better than the audio/video from the Techstars.tv feed.
Next time, there should be scheduled time slots for applicants to pitch to mentors; everybody learns from listening in to the others. Since San Antonio will be my first Techstars, I picked up context and background which will help me be more useful more quickly in the early January sessions.
Since Dave Ward's presentation on Software Defined Networks at Nanog there's been a lot more of it about. SDN feels a lot like MPLS (Multi-protocol label switching) did when it was first introduced - it appears to simplify the operation of the network without actually doing so, and it gives the industry a 'big new thing' to talk about and compete over.
Ivan Pepelnjak thinks so too - he's enthusiastic. So is James Hamilton (Amazon).
Dave Ward (now Juniper) did the first day keynote on Software Defined Networks. The proposition is that it should be possible to give a greatly improved experience to users if it was easier for applications and the network to exchange information. Protocols: ALTO, BGP-TE, PCE, Openflow, Services Registry. Randy Bush observed in the Q&A that this kind of thing has been tried before, several times, and that security and privacy conflict with it.
Greg Hankins (Brocade) reviewed current router architecture requirements. Line rate 100 GbE is one packet every 6.72 nanoseconds. The only memory technology that supports this is embedded on custom ASICs - transfer rates to SRAM or DRAM chips are too slow. 3D packaging might help, but it generates heat in even more density than the existing 2D components. There's a useful table showing what can fit on an ASIC at each process geometry up to 32 nm (slide 14).
Peter Hoose (Facebook) gave a lightning talk about troubleshooting - it hasn't changed all that much in the last 20 years, although networks have evolved hugely - he doesn't have solutions, but there's an opportunity here.
Rob Sherwood gave the Tuesday keynote about the OpenFlow abstraction - there are at least 75 public implementations, a standards body, and demonstrated interoperability. Use cases in presention - a virtualized control plane; a VM moved from Stanford to Japan without changing its IP address; rerouting of flows between WiFi and Wi-Max; graphing live power and latency variation in a data center.
Todd Underwood (Google) was hugely entertaining in his lighting talk about reliable use of probabilistic IPv4 prefixes as a mitigation for the runout of IPv4 addresses. Great acronyms !
Geoff Huston was pessimistic about the liklihood of a coherent single Internet persisting as the IPv6 transition happens at different rates in different geographies.
Matthew Petach (who wasn't at the meeting either) took great verbatim notes from the live video stream. If this event is like previous Nanogs, eventually the video will be posted on the agenda page against each agenda item. Meantime pdfs of the slides for each presenter can be found there.
The Valley is a strange place most of the time - there's a willing suspension of disbelief which allows for optimism, so people start companies and get financing for unlikely ideas.
Yesterday, though, it was the reality that was strange. It started with a posting on Facebook from friends who have a house on Lorne Way, in Cupertino - there was a manhunt in progress, for someone who'd shot several co-workers earlier that morning at the big cement quarry a few miles to the West. Some of the comments, about the advisability of obtaining and using one's own guns, were interesting. #Cupertino was a trending tag on Twitter, at least for the San Francisco area - very little real content, lots of people repeating each other.
In the afternoon, Apple announced that Steve Jobs had died - the day after the big presentations for the launch of the iPhone 4S. That created another flood of reaction. There was rain and intermittent sunshine, producing rainbows, which were declared 'a tribute to Steve'. As ever, Apple's information control was impressive - Tim Cook, Apple CEO since August, doing his first launch as CEO, must have known that Steve Jobs was at home - but nothing leaked, from either Apple or the medical people.
This morning (6th October) the "suspect in the shooting spree" was shot and killed.
People who have had a little time to think and reflect are writing about their interactions with Apple - here are a couple of more personal reflections from people about their early development experiences.
NESTA is the National Endowment for Science, Technology and the Arts - an independent body with a mission to make the UK more innovative - it is funded by an endowment from the UK National Lottery. It makes investments, and also publishes research.
Yesterday's blog post was notes about a visit to PlugandPlay Tech Center - which is primarily a place for startups to rent office space on extremely flexible terms; it also offers shared space, access to advice, and other things which might help startups get going more quickly (or fail an idea more quickly). Either way, this usually saves time and money compared to doing it alone. PlugandPlay is one example of an incubator. Y combinator, of course, is another. NESTA doesn't mention either of these big examples in its recent publication, focusing instead on Cambridge, England, and on Sweden for examples of success. From the perspective of the Valley, it seems odd to ignore these big operations.
out of an Acorn ...
Nevertheless, here are some useful quotes from the Incubation for Growth paper.
"Absolute measures of incubation are impractical, but performance indicators are useful"
" Business models have changed over time. Since 2005 there has been a reported increase in the cost per job created each year in business incubation in Europe. Some incubators now offer equity finance, and some equity investors offer incubation, with an unclear distinction between both. The challenge for incubators and their funding bodies is to capture some of the value created for incubatees. Generating revenue from services when clients are resource constrained is often not possible without subsidies from public bodies. Corporate funded incubators typically require a strong strategic fit of incubatees with the corporation, which is not appropriate for all ventures. Incubators with mixed funding may encounter principal-agentproblems as they attempt to meet multiple objectives."
"Longer term, the challenge for incubators is to help develop a relatively small number of high-growth,
internationally competitive companies, even if these companies realise their success after incubation."
"Here incubation diverges from venture funding. Unlike finance managers reliant on realizing an investment portfolio at a significant profit, incubator managers have several sources of income – rent, grants, public or private consulting, conferencing and other services, share options – and so also numerous constituencies to please, whose interests may not always be obviously reconcilable. Part of the craft of the incubator management team is to understand such conflicts and resolve them so far as possible, while retaining ‘negative capability’ – an ability to improvise and live with the doubts and uncertainties inherent in an entrepreneurial environment."
This lengthy report assumes that an incubator is primarily focussed on real estate, renting out space to startups, and that it is therefor geographically anchored to one place.
It is possible that other models, building on the 'tech team in one place, sales and product marketing in the large marketplace' principle, which works for startups from small countries, could work.
3 March 2012 adding reference to Nesta Startup Factories list
29 June 2012 update to add Forbes top ten list of incubators (there's still no agreement about whether these things are incubators or accelerators)
26 July 2012 update to link to Jed Christiansen's list of accelerators http://www.seed-db.com/accelerators
Tom Willey and I went along to the Expo at the Wolfe Rd facility last week. It was very crowded, maybe 400 people in a space which'd have been comfortable for 250. Lots of enthusiastic young companies, and a few people with more experience.
Having watched 20 or so of the 32 presenting companies, here are notes.
Zeromail reminded me of Microsoft Outlook, about 3 versions back, before it was completely overtaken by featuritis. http://zeromail.com
MediaFunnel is a 'dashboard' for enterprises who need to manage a social media presence over many properties. Use it to find what is being said about the brand, as well as to coordinate a public presence. http://mediafunnel.com
Realglobe Inc run a cloud hosting service called C4SA in Japan - they aim to sell it in the US too. http://www.realglobe.jp
Narvalous is building a social game - avatars, 3D and all.
Carbon Lighthouse will take on energy management for your buildings to save you money and emit less carbon. This would make more sense in geographies with a functioning carbon tax regime. http://www.carbonlighthouse.com
eThor (not a useful name) has built software to connect mobile end systems to retailer's Point Of Sale systems - the example was food orders, but it's broader than that. http://ethorlink.com
Mobilitas SpA have built a platform for designing location-based mobile games - from Santiago, Chile
Snackr is open source code to create an alternative display for things in your RSS feed. I have all I can do to get through the updates for 400+ items in my feed, without having random items from it scrolling on my desktop. Wonder what they are really building ? http://snackr.net
Summary - some real products, lots of things which have been done before, many things which hadn't settled on a revenue model. Glad I wasn't judging, though there was a list of 70 companies, investors and big corporations, at least half of whom had actually sent someone to judge, from the look of the crowd at the front of the room.