Startup Engineering Cookbook

Are you a non-technical founder confused about the technology to deploy for your new website? The presentation covers the various options available for building a website and beyond...

Transcript

1. June 2013 Startup Engineering Cookbook From Slides to Website
2. Overview Slide Topic 3 Purpose 4 The Lean Startup 5 Implementation Options 6-17 Architecture Options 18-22 Sample Architecture 23-27 Monitoring 28-29 Disaster Recovery 30-31 Scaling 32-33 Internal Tools 34 The End 35 Contributors
3. Purpose Covering the various technical options when going from a concept to building a website and beyond... 3 [meant for a non-technical audience]
4. The Lean Startup It’s an approach for launching businesses and products, that relies on validated learning, scientific experimentation, and iterative product releases to shorten product development cycles, measure progress, and gain valuable customer feedback. [ Wikipedia ] It is THE industry bible for startups...READ IT. [ The Reality ] 4
5. Implementation Options There are 3 options for building a website: 1 Build internally 2 Outsource 3 Hybrid - hire a technical lead Building a website is like the Kama Sutra... many ways to accomplish “it” 5
6. Architecture Options 6
7. Infrastructure Provider The website will have to reside on a computer for the world to see. Options: Amazon Web Services (AWS), Google Cloud Platform, Heroku, local hosting provider, Microsoft Azure, Netmagic, RackSpace Our choice: Amazon Web Services (AWS). They are the #1 leader in web infrastructure services also known as cloud services. In addition, they have a free 1 year trial which allows you to try many of their services at zero cost. After 1 year they will bill you based on usage of their services. web hosting services 7
8. Central vs. Distributed The old way of designing the infrastructure for a website was to put all the resources such as databases, files, images, etc... on the same server. The problem is if your database is slow and everything else is fine, you have to upgrade the entire server and disrupt the other functions. By distributing the resources you can quickly pinpoint the bottleneck and scale that particular resource. distributed web architecture Central Our choice: Distributed Architecture. Distributed Computing Database Static Files Computing, database, files, etc... 8
9. Managed vs. Self-Service If your team is technical it might not be technical in terms of hardware. Just because someone knows a programming language does not mean they have the infrastructure knowledge to run a website. Some startups prefer to outsource the management of their hardware to a hosting provider or use the “premium” option of a cloud service provider. Our choice: Self-Service. The tools that are provided by AWS are quite powerful and easy to use for managing the entire AWS infrastructure. managed server hosting 9
10. Operating System There are only two options Microsoft Windows or Linux. Windows is limited in the type of programming languages that you can run on the server. Linux is open source and has a large community that is continually improving the operating system. Within Linux there are many options such as Debian and Red Hat to choose from, but are all pretty similar in their offerings. Our choice: Linux. Linux is usually cheaper to run and used by more websites than any other platform. Part of the LAMP stack - Linux (operating system), Apache (web server), MySQL (database) and PHP (programming language). linux hosting 10
11. Software that delivers web pages from the server to a clients web browser. Options: Apache, Lighttpd, Microsoft Internet Information Server (IIS), Nginx, Node.js, Rails, Tornado Our choice: Apache. It’s part of the LAMP stack and is widely used by many websites. Web Server apache web server 11
12. The heart of any web based application is the database since it stores all the user information and everything else related to the web application. Options: Amazon DynamoDB, Cassandra, CouchDB, Microsoft SQL, MongoDB, MySQL, Oracle, SimpleDB Our choice: MySQL. It’s part of the LAMP stack and widely used by many websites. Database mysql nosql 12
13. If you ask 4 programmers which is the best language, you will most likely get 5 different answers. If you are building a website on a Windows environment then you can choose .NET, for Linux there are several choices. Options: Clojure, Microsoft’s .NET, Perl, PHP, Python, Ruby Our choice: PHP. It’s part of the LAMP stack and more importantly has a large community so finding people to hire with PHP knowledge is better. Back End Language web php languages 13
14. Instead of re-inventing the wheel it’s better to use a pre-existing framework to start building an application. Frameworks provide a foundation where developers’ code can be neatly arranged. Also, documentation, common problems, solutions, tips and advice can be shared online in an open source manner. Options: CakePHP, CodeIgniter, Laravel, Symfony, Yii Framework, Zend Framework Our choice: CodeIgniter. One of the fastest performing frameworks and very active support community. PHP Framework php framework 14
15. This is what the consumer sees on their phone, tablet or computer screen. Previously, the trend was to create a website design that was different for each type of device that was accessing the site. Now everyone is moving towards a “responsive” design, which means you design a single site that can be viewed on any device that a consumer may own. Options: Bootstrap, Foundation, jQuery UI, Kendo UI Our choice: Bootstrap. It was developed by Twitter and released as an open source project which means there is a large community of users. Front End UI Framework front end framework 15
16. Version control is a system that records changes to a file or set of files over time so that you can recall specific versions later. It’s a service that will make your life easier as you will be able to keep track of the changes to your code base. Options: BeanStalk, Bitbucket, GitHub, ProjectLocker, Springloops Our choice: Bitbucket. They have a free plan that allows 5 users which should accommodate any startup. Version Control Service git version control service 16
17. You can interact with your online version control service using the command line interface of your computer. However, many people prefer to use a graphical front end which is usually referred to as a git client. Options: Gitbox, GitHub, SourceTree, Tower, Versions Our choice: SourceTree. They have a Mac and Windows client which are both free. The software is from the same people that provide the Bitbucket service, so they very well together. Version Control Client git client Tower Versions 17
18. Sample Architecture 18
19. UI is a small piece 19
20. Day 0 Elastic Compute Cloud (EC2) t1.micro RDS MySQL t1.micro S3 Region: APAC- Singapore Single Region, Single Availability Zone (AWS Free Tier Program) Users Internet 20 Availability Zone: AP-Southeast-1a
21. Day 0+X RDS MySQL S3 Single Region, Multi Availability Zones The fault tolerant design allows for one of the availability zones not being available. Users Internet Database Replication Elastic Compute Cloud (EC2) RDS MySQL Availability Zone: AP-Southeast-1b Elastic Load Balancer (ELB) S3 Elastic Compute Cloud (EC2) 21 Region: APAC- Singapore Availability Zone: AP-Southeast-1a
22. RDS MySQL S3 Users Internet RDS MySQL Elastic Load Balancer (ELB) S3 Day 0+2X Multi Region The fault tolerant design allows for one of the regions not being available. The design also allows for users to be load balanced based on geographic location. Availability Zone: US-East-1a Elastic Compute Cloud (EC2) Elastic Compute Cloud (EC2) 22 Region: APAC- Singapore Availability Zone: AP-Southeast-1a Region: US East (N. Virginia)
23. Monitoring 23
24. By monitoring the infrastructure, you can proactively plan if certain components need to be upgraded. Infrastructure Performance 24
25. Understanding how the application affects the infrastructure is essential. If there are very few users but the servers and databases are consistently running at peak utilization, that might indicate a programming issue. Application Performance 25
26. Google Analytics is the industry standard for tracking website statistics such as hits, users, content viewed, conversions and many other statistics. Traffic Analysis 26
27. Once a user starts to use your web application you will want to engage with the customer. Intercom.io allows you to send emails based on pre-defined rules such as “send a welcome email to a new user”. User Engagement 27
28. Disaster Recovery 28
29. Having a disaster recovery plan in place is critical to ensure the website can recover from a catastrophic event. Taking a daily/weekly/monthly snapshot of the servers, files and databases is a minimum requirement. The daily/weekly/monthly backup schedule is sometimes referred to as the son/father/ grandfather rotation schedule. Backing Up & Restoring If disaster does strikes and a complete backup strategy is in place, then it should not be very difficult to restore the website from the backups that are being maintained. 29
30. Scaling 30
31. Hopefully, the website you launch is a massive success and leads to your website having a tough time handling the load. With many cloud service providers you can automatically add additional resources based on predefined rules. An example is below: Adding Resources Rule Action If CPU utilization goes above 60% during a 5 minute period Upgrade the server to the next larger size to handle the additional load 31
32. Internal Tools 32
33. Team communication is critical in a fast paced web business. Yammer is an internal social network for employees. Pivotal Tracker is a project management tool. HipChat is an instant messaging application. Team Communications 33
34. The End 34
35. Sahil Parikh Founder Brightpod @sahilparikh Contributors Manish Jain Co-Founder MProfit @mrjain mrjain@gmail.com Kulin Shah Co-Founder Wishberg @NowEntrepreneur 35 Pankaj Jain Venture Partner 500 Startups @pjain