Heroku: API Unavailable

I went to update some content on my Heroku hosted Python/Flask app and was unable to upload / git push the app due to an "internal server error" and associated "We're investigating issues with an API database failure. We're recovering the database now" message.

Pretty scary, and not a confidence builder given it's my first week testing the service. However. on the upside the downtime (for me) was just a few minutes and the live production web app was still available during this time using the public URL.

Below is a log of the session.

Login... Password (typing will be hidden): ! Internal server error. ! Run `heroku status` to check for known platform issues. j2mac:jello j2$ heroku status === Heroku Status Development: yellow Production: No known issues at this time. === API Unavailable 2013/02/28 19:27:56 (~ 13s+) 2013/02/28 19:27:56 (~ 13s ago) issue We're investigating issues with an API database failure. We're recovering the database now. j2mac:jello j2$ heroku status === Heroku Status Development: yellow Production: No known issues at this time. === API Unavailable 2013/02/28 19:27:56 (~ 1m+) 2013/02/28 19:27:56 (~ 1m ago) issue We're investigating issues with an API database failure. We're recovering the database now. j2mac:jello j2$ date Thu Feb 28 11:29:13 PST 2013 j2mac:jello j2$ git push heroku Counting objects: 7, done. Delta compression using up to 8 threads. Compressing objects: 100% (4/4), done. Writing objects: 100% (4/4), 1.06 KiB, done. Total 4 (delta 2), reused 0 (delta 0) -----> Python app detected -----> No runtime.txt provided; assuming python-2.7.3. -----> Using Python runtime (python-2.7.3) -----> Installing dependencies using Pip (1.2.1) Cleaning up... -----> Discovering process types Procfile declares types -> web -----> Compiled slug size: 23.9MB -----> Launching... done, v77 http://stormy-citadel-6634.herokuapp.com deployed to Heroku To git@heroku.com:stormy-citadel-6634.git 8563d23..0fb61d6 master -> master j2mac:jello j2$

Idealist (sortable): Python, Flask, Jinja2, Javascript and Tablesorter

I've been experimenting with Python and the Flask web development framework. Combined with hosting on Heroku, I was surprised how quickly one could get a web app up and running. For my experiment this weekend, I created a sortable table of the business model and technology ideas from my previous blog post.

In the spirit of documenting my efforts, here are the ingredients to this micro app:

• Python v2.7.1: An interpreted open source programming language, with an incredible number of modules/resources available
• Flask: a Python-based web development language (BSD Liscense)
• Jinja2: a Python-based templating engine for web development
• Heroku: a cloud-based web application hosting platform

Below, is a sample of the result embedded from the live Heroku site, but since this Blogger blog platform seems to scrub Javascript from the iframe to be able to sort the list may have to go directly to the site hosted at Heroku.

<Note: this may take some time to load as it's hosted on a development Heroku dyno and goes into sleep mode when accessed sparsely.)

FBI Surveillance Van - WiFi Network

I've heard of people having fun with naming their WiFi networks with funny names and the FBI strategy is always mentioned. So I was amused this week as I was out to dinner with the kids in downtown Santa Barbara. I pulled out my iPhone and this is the network list that was found. Pretty funny to see the "FBI Surveillance Van" name used in the wild. While my first thought was, "ha ha, that's pretty funny," I must admit, I did have a gander out the window to see if there happend to be a blacked out van parked out front.

Nike+ Accelerator Ideas

On December 12, 2012, Nike announced their Nike+ Accelerator program in which 10 companies, selected by way of an online application, will be given API access to Nike+ API, mentoring, $20k, and several months of time working at the Nike campus in Portland, Oregon. The program is for companies that intend to leverage "Nike+ technology to create products and services that will inspire athletes across a broad range of activity and health goals including training, coaching, gaming, data visualization and quantified self." The programs is managed in conjunction with TechStars.

As an athlete, mobile app developer and Internet of Things enthusiast a friend who also happens to be a mentor for the program suggested I have a look. While I opted not to apply for various reasons, I thought it was interesting to brainstorm a couple of concepts, just to see what popped.

Data Access & API
First off the data & API, the opportunities are really dependent on the kind of access is available.  What data available via the API provided by Nike?

The information on the Nike developer site are limited but it appears that the data available on the site is limited to the data for a particular user. For example, I could build an app in which a user could log in and access their Nike+ data, which would allow me to present in the app the user's data in my own format or for my own purposes. 
Nike describes their data access into the following four categories. I have included the JSON (JavaScript Object Notation) data structures at the end of this post. These are directly from the Nike developer documentation. 

Aggregate Sports Data: This is a list of "Nike experiences" for which the user has data. An "experience" is, I believe a category of data. For example the Nike running app would collect one category of data and their Fuelband product would collect another.

List Activities: Lits of activities with with summary information for each, including Activity ID, time, duration, calories burned, NikeFuel earned and the device which recorded the data.

Activity Detail: Provides similar data attributes as the data structure above, but for an individual activity.

GPS Data: GPS data in 3-dimensions (lat, long & elevation) for an activity.

Accelerator Questions - Nike+ Accelerator / TechStars

I was looking at the Nike+ Accelerator program recently and found that I had referenced the submission questionnaire a couple of times. I thought it was worth summarizing the questions here as they are part of a general question format (from the Unified Seed Accelerator Application by the Kauffman Foundation and TechStars) that various accelerator programs use. While perhaps the jury is still out on how effective these accelerators will be in building sustainable companies, the questions are good. Each of the various programs custom tailor their own questions, but they seem to mostly use the same ones. 

Below are the questions from the Nike program.

• What is the name of your company?
• Please describe your business in 140 characters or less.
• In more detail, what will your company do or make? What's new, interesting, or different about what your company will do?
• If you have a website or demo/prototype, what's the URL? Please provide username and password if necessary.
• Business Video: Please provide a video < 3 minutes long which best describes your business. Production value is not important to us, it can be quick and dirty.
• Team Video
• Please provide a video < 3 minutes long which best describes your team. Production value is not important to us, it can be quick and dirty.
• Please tell us about each founder and their role. Explain how you met and how long you've been working together. Please be thorough because we place a great deal of importance on the team.
• For each founder: First Name, Last Name, Email, Role/Title, LinkedIn Profile, Twitter, Github Profile
• Explain how the company will make money.
• Please provide information on current or likely competitors. Include key differentiators. Please include URLs.
• What are some things that the team (or its members) have built in the past? Please include URLs.
• Can each of the founders attend the entirety of the program, or do some of you have other obligations during the timeframe of the program? If not, please explain.
• Not including the founders, how many additional employees are there? Please provide LinkedIn profiles and Github URLs (if applicable).
• Where is your company located?
• Please provide information on money the company has already raised, and any information on fundraising plans for the future.
• Why should we choose your company?
• I understand that Nike will receive many applications for this program and many of them may include similar ideas. I also understand that Nike is under no restriction with regard to the companies it contracts with and the products and services that it develops or offers.

The Killer (Sensor) Surfing App

I had sensors on the mind after my post documenting some of my favorite mobile device sensors, so when a friend said there was still some surf at Rincon, I decided to head down this morning. The surf dropped since yesterday, but was still fun. However with sensors on the mind it really got me about the ultimate sensor. The killer app for sensors. That would be a sensor that lets you know that the surf is good. 

But before you say, hey just go look at the models. That's not exactly the point. Kind of, but not exactly. I love SwellWatch 3D - here's the Rincon forecast and I always appreciate that resource. But the problem with the forecast is that it only shows a couple attributes. These are key attributes, yes, but still missing a couple elements that characterize a good session. Surfing is just incredibly fickle, and so many minor factors influence a good session. For example, today the model looked a lot like the day before, however, if you'll notice it's about shoulder high, whereas the day before it was head high, and way more consistent. Also, it was more crowded than yesterday. If you look in the background of the pictures, count the surfers.... a lot. And that's only a small slice of the point.

Mobile Device Sensors

I've been evaluating the sensors available for the various mobile devices and thought I'd document some of the highlights. Sensors are an interesting aspect of mobile devices because your mobile device, whether Android or iOS,  tablet or phone, solve the last mile communication problem of many sensors. Many sensors are very local in their communication distance capabilities, either tethered with a cable, or short range like the 50 meter Bluetooth Low Energy range. However, your mobile device can link a sensor to the Internet via your wireless cell phone carrier, or via your phones WiFi connection. Furthermore mobile devices provide the computing platform and user IO functionality for sensors. So with mobile devices you have the computational, communication and user IO covered, which allows for cost effective development of sensors and sensor applications. I'm interested in sensors for both end-user capabilities as well as APIs so I'll be sure to note any APIs.

For starters, similar to the my business model and technology idea list, here is a high-level overview of some of the sensors I've been finding. 

External Sensors

HiJack: More of a sensor development platform than a sensor itself, Hijack is a University of Michigan project that provides a general purpose hardware and software platform to build sensors using power and IO from the iPhone's 3.5mm iPhone audio jack. The hardware, pictured below, is a small board with an integrated headphone jack. The software API allows the creation of general purpose sensors. The hardware is available at Seeed Studio. Some demo project built on the platform include an EKG monitor and soil moisture sensor.

The Idea Funnel

If you envision the process of brainstorming an idea to actually executing on a business opportunity like a sales funnel, then you might break up the process into distinct phases:

Phase 1 - Brainstorming

Phase 2 - Filtering / Reducing

Phase 3 - Market Validation

Phase 1 is all about volume, just brainstorm ideas. Then Phase 3 is where you might go after 1 idea or maybe do an initial market validation exercise with perhaps 3 ideas. Phase 2, and what I want to explore here, is how you narrow the ideas down from a bunch to just 1 to 3 ideas? I am not sure, but the first step might be to determine the criteria. These will be different for different people or organizations, but for me here is what comes to mind:

Business Model and Technology Ideas

As a believer that it’s not about the idea but the execution, and further that even a good idea will really evolve as a team vectors into the right set of features and market opportunity, I thought it would be safe (and fun) to list some of the various ideas I’ve looked at recently. Some of these are based on a need that I have (but not necessarily anyone else). Others are just random ideas. Some just look like fun to build. And some I guess are just downright ridiculous. But in the spirit of brainstorming, where more is better and it’s better to just get some ideas out there to evaluate, reduce and solicit feedback, I include the list below. Feedback welcome:)

I have also experimented with a sortable version of this idealist using Python/Flask/Javascript/tablesorter and hosted on Heroku.

Approach to Market Validation

I was recently talking to a California company about moving their products into the enterprise security space. The company has some really interesting data analytics capabilities that correlate complex relationships in big data. Below was my first go at documenting an approach for how I might attack the effort. I've been a student of the market validation process, and - to give credit where credit is due - I have included references below. Also, thanks to David Telleen-Lawton (and his 20 years of market validation experience) for feedback on this.

Overview of Market Validation Approach
Market validation is a structured process in which new product offerings are validated in terms of customer need and product requirements. The process provides a rapid, structured and repeatable approach to vetting ideas, identifying potential customers, market segments and messaging, while quickly determining the Minimum Viable Product (MVP) feature-set such that a viable product, business model and sales approach can be quickly identified. The antithesis of this approach is to define requirements in a vacuum after unstructured feedback from 5 meetings, spec out an expensive v1.0 product release, build it for a year only to find out that you built something that nobody wants. Or talking to 10 customers and identifying 10 “must have features” and building a product that only 10 customers will buy. The market validation thinking is more about quickly getting customer feedback and iterating the product (back of the napkin, diagram/presentation, mocked-up demo, prototype, beta version), such that when version 1.0 is released you know you are solving a generalized problem that customers will pay for - and you will have tested the messaging, know the sales requirements/channels and have tested the pricing. In fact, ultimately, you’ll already have your first paying customers lined up before you release v1.0. The process as outlined below (the timeline is measured in weeks, or a couple of months) begins this iterative process and defines the product, customers, problem and product-market fit, but the process should really continue into perpetuity as the product continually evolves. If the real estate mantra is location, location, location, then the market validation mantra is, get-in-front-of-customers, get-in-front-of-customers, get-in-front-of-customers!

Exercise to earn credit to watch TV

As I evaluate opportunities, one idea that I have been looking at involves apps, hardware and fitness for students, among other things. In the spirit of "it's not the idea, it's all about the execution" I felt no risk in posting an overview of the concept here. As a mobile app developer I really understand the app dev cycle, but the hardware component here is unchartered territory for me, though I believe there may be some opportunities to leverage an existing bluetooth power strip and API for phase one.

Exercise to earn credit to watch TV

The plan ties into the craze for fitness apps and sensor devices, such as Fitbit, Nike+, but leverages all of that rather than competes. It scales and can be built iteratively. Furthermore, the platform could be extended to address various markets, so the market potential is significant. 

The concept is to create a system that requires kids to exercise to earn credits which would be redeemed to watch TV. This ties into the American collective consciousness that kids are not getting enough exercise, watching too much TV, gaining weight, getting diabetes, etc., and generally increasing the burden on the US healthcare system ($2.5 trillion dollar industry and 17+% of GDP*). The credit system addresses a problem of such scale that it is a key initiative of Michelle Obama** and UnitedHealthCare*** .

Phase 1: Create an app and associated hardware device (bluetooth enabled smart power strip). The app stores credits and those credits are used to turn the power strip on from the app. The TV is connected to the power strip (you could create a locking version to prevent tampering, or just include a unique colored zip tie or security tape that would make it obvious if the power strip were bypassed). This could be live in perhaps 6 months. The credits could be input into the device manually by a parent for starters, but could quickly be extended to include the sensors in the phone, other apps or a Nike+ device.