Fedora Cloud Product Requirements Document.
Document Purpose and Overview
What this document describes
This is the wikipedia Product Requirements Document for the Fedora Cloud SIG. It:
- Provides a high-level market of the cloud computing market as it pertains to the Fedora Cloud SIG; this includes overviews of things which may not be within our actual scope/ability to accomplish at the current time.
- Provides deeper understanding of the types of users who could use Fedora for their cloud computing needs. This includes describing their main day-to-day tasks, common problems, etc. The perspective here is not necessarily limited to system administrators, or developers, but a combination of many types of users and roles.
- Ties common issues and needs of potential users/consumers of the Fedora Cloud product to high-level product needs, from a "functional" standpoint
- Provides solutions, in the form of "changes" or "features" which will provide the functionality described as needs for the potential users.
Fedora Cloud provides a customizable base image and tools for developing scale out applications on public and private clouds, as well as a small number of images pre-configured for specific uses. Public and private cloud adoption is taking off, and the requirements for an image OS differ significantly from the requirements for a desktop or server OS. In these environments, much or all of the instance lifecycle — from the creation of the image and addition of software or configuration specific to the instance, to the teardown of the image — will be automated. Systems are designed to scale out via many identical nodes rather than scale up with carefully-tended individual servers. Individual uptime (mean time between failure) is not as important as the ability to get a new instance running quickly (mean time to recovery). With that in mind, we're tailoring a release specifically for cloud environments.
Public and private cloud adoption is happening rapidly, but the market is not yet mature and is relatively ripe for disruption even though some favorites have emerged as early leaders. In the next two to three years, we expect to see a great deal of growth in adoption and still see a number of emerging players where no clear favorites have emerged (for instance, Google Compute Engine). Additionally, some platforms (Amazon Web Services) have matured to a point where a large number of companies are relying upon the technology for their full infrastructure. While this is not a widespread practice, AWS is seeing a great deal of adoption and will likely start eating into "traditional" workloads that currently live behind the firewall. In short, there's an enormous opportunity for Fedora to become an instance-OS of choice if the project moves quickly, develops or adopts the right technologies, and succeeds in educating the market about its existence. A failure on any of those three points means that the Fedora Cloud product will have little chance in taking a significant portion of the new market or taking any of the existing market.
The Fedora Cloud product consists of an image to be used to run one or more instances in a public cloud, and a set of tools for creating and modifying images. We will also provide two to four additional images that are pre-configured for what we expect to be the most popular scenarios/use cases for using Fedora in public or private cloud.
Major Release Themes
Cloud computing in general is the transition of computing power from individual hand-tended resources to a ubiquitous utility. Fedora fits into this at several levels, from the infrastructure service software we include (like OpenStack and Eucalyptus) to end-user tools. The Fedora Cloud image fits into middle of this, providing a guest OS image to run on Infrastructure as a Service systems, on which platform and application services can be deployed. We targets use cases which fit the "cattle" side of the "pets vs. cattle" metaphor for computing.
Aside from adoption and development of applications on top of the Fedora Cloud images, we have a few secondary goals that should be helped by wider adoption:
- More testing of Fedora images with additional bug reports.
- Better feedback about how the product should improve. This is separate from "bug reports" in that we hope to engage the audience and receive detailed feedback about use cases, desired features, developing trends in cloud management, etc.
- Patches and contributions that will help improve the product, and Fedora in general. Assuming we're successful, some users should take an interest in helping to develop our product.
Target Market / Audience
Developers creating scale out applications on top of public and private clouds, and organizations and users running those applications.
We need to ensure that images are as easy to consume as possible. Since Fedora Cloud images are meant to be consumed as part of a public or private cloud, we won't be worrying about physical media at all. The cloud images also won't be "installed" in the same manner that users are accustomed to with desktop or server images. The cloud image will simply boot in its target environment ready to run and/or for further customization/configuration.
Where to obtain
Users will be able to obtain the images for public clouds via download or via the usual marketplace for those images. For instance, we publish Amazon AMIs on Amazon directly. Users are able to launch new instances with Fedora without having to obtain the images directly from the Fedora Project and then upload to Amazon. Users will be able to download appropriate images for Apache CloudStack, Eucalyptus, OpenStack, and other IaaS platforms.
Images will be delivered as AMIs on Amazon EC2, and as downloadable images in qcow2 and raw.xz formats. We may add other public cloud images and other downloadable formats to meet demand or anticipated need.
Fedora Cloud images can be updated using yum as normal. We also intend to produce periodic respun images with security updates, once the infrastructure is in place to support that.
Image Creation Toolkit
We will also maintain a set of tools that can be used to generate, modify, and configure Fedora instances for use with public and private clouds.
Currently, Fedora is not a widely used option for instances on public and private clouds. We know there's some usage, but it's not one of the top three or four OSes on Amazon or (likely) for private clouds. Success looks like:
- Increase in adoption.
- Third party support / targeting of Fedora Cloud as a platform.
- Increased contribution and participation in the Fedora Cloud WG and Fedora Project in general.
User Profiles, Goals, and Primary Use Cases
Still working out some logic on this section. But forging ahead with what I have for the moment. Goal of this section is to provide insight into either or both of:
- Primary Use Cases: What are the situations / environments in which we expect a Fedora Cloud Product to be used
- User Profiles and Goals: This is more like “personas” work, or could be done as “user stories” (more along the lines of agile, see http://en.wikipedia.org/wiki/User_story )
... and then ensure that for each type of user or use case, we have features/changes that make the Fedora Cloud product useful.
Based on OpenStack personas (licensed under CC-By 3.0). Three Cloud User Roles (based on “Description and Application of Core Cloud User Roles” ACM CHIMIT 2011, December 4 2011) that describe the tasks of the people who interact with any cloud-based Information Technology system:
- Cloud Service Creator: Develop the technical and business aspects of a (simple or high-level) cloud service
- Cloud Service Provider: Provide all types of services (SPI, etc.) to a Service Consumer
- Cloud Service Consumer: Consume all types of services (SPI) offered by a Service Provider
Persona #1: Alan the AWS enthusiast
Alan was a very early AWS adopter. He writes and maintains a number of AWS-deployed applications, including staging and production, from the same data source. He needs to monitor his applications for cost, resource consumption, demand peaks and any crashes. He works for a small start-up. Puppet master.
- Adoption curve: Innovator
- Skills: Experienced web developer & operations guy - not on his first cloud application architecture, mastery of continuous integration (CI) and continuous delivery (CD).
- Behavior patterns: Wants to concentrate on application development, not architecture. Change management - applying security updates to guest systems - takes a chunk of his time. He spends a little time every few months evaluating the cloud architecture to consider alternative components to increase price/performance.
- Goals: To deliver excellent scalable web-deployed applications with a web interface and robust API for a mobile application front-end. To deliver incremental, tested updates to the applications on a regular basis.
- Needs: Application lifecycle management, monitoring, to make his apps faster. Makes heavy use of cloud APIs and services to minimize work he has to do setting up his own services.
- Main Tasks: writing and deploying code. Anything that requires a significant distraction from that is a time sink, and Alan has little tolerance for mucking with infrastructure more than he has to.
- Attitudes: Alan loves AWS. He's unlikely to consider an alternative public cloud due to dependencies on AWS services and the fact that migration would consume too many cycles that he doesn't have.
- Beliefs: Continuous deployment is king. Automate all the things. Do the hardest things often, until they’re easy. etc.
- Motivations: getting the code running in production
- Frustrations: poor command line tools,
- Environment: Macbook for development, AWS for development, testing, and production environments.
- Interface Usage Tendencies:
- GUI: Medium
- CLI: Medium
- API: High
- Collaborates With: Alan is pretty much working alone. He may work with other folks in the start-up, but not on a daily or regular basis. His work flow is tooled specifically for his convenience.
Persona #2: Erin the Rails devops team member
Erin uses the cloud to do Ruby on Rails development utilizing virtual machines.
- Adoption curve: Early adopter
- Skills: Familiar with Linux systems. Several years' programming experience, comfortable with Linux and other environments.
- Behavior patterns: Will choose from the range of available flavors of VMs to run. Will use OpenStack images to launch instances, also create and save her own.
- Goals: To set up and use cloud guests for her application with minimal effort. Erin is focussed on the application and regards the IaaS services as resources to be used with a minimum of administrative overhead. Would prefer if her language stack were just there and just worked in the cloud guests of her choice.
- Needs: Easy way to create cloud guests configured with the versions of Ruby and Rails used for her various projects.
- Main Tasks: Deploying and updating code in development, testing, and production environments. Maintains the guest images used as a base in all three.
- Attitudes: Hates it when things change just for change's sake. Wants the latest of the things her team needs, wants the rest
- Beliefs: The development, test, and production enviroments need to be identical.
- Motivations: provisionning environments quickly and easily
- Frustrations: lacking guest images preconfigured with her application framework
- Workflow: Will mostly use the GUI and command line to set up and interact with OpenStack services, and will ensure that rolling out the application will be automated as much as possible. She will use the dashboard to monitor consumption of resources on an ongoing basis. Some else in her group has primary responsibility for managing the IaaS layer, but she can pitch in with a basic understanding if need be.
- Environment: Linux development system. Primarily launches cloud instances via the OpenStack GUI. Part of a devops team which manages the a small OpenStack deployment for their own use.
- Interface Usage Tendencies:
- GUI: Medium-High
- CLI: Medium
- API: Low
- Collaborates With: Other members of her development and production operations team.
Persona #3: Walter the Web Developer
Walter develops feature rich HTML5 applications using Symfony and Twitter Bootstrap, and a range of other web technologies. He has no real system administration skills, and develops on a single desktop computer or his MacBook. Walter wants the app to Just Work when he’s finished, and doesn’t really think a lot about caching, sharding, proxies, or making his app scalable. He works in a medium sized development shop which has people who take care of deployment and change management to the web applications he produces. By-line: “Not a bug: Works for me”
- Adoption curve: Early majority (Early adopter for web technologies)
- Behavior patterns: Wants to concentrate on application development, not architecture. Spends most of his time in TextMate, and testing how app behaves and looks in VirtualBox.
- Goals: To deliver excellent web applications and robust RESTful API for mobile apps.
- Needs: Up to date application development stack, good developer tools, someone to actually deploy his app.
- Main Tasks:
- Attitudes: Walter would like to know more about the cloud but there’s so much to learn, he has no real idea where to start, and just keeping on top of new web development trends takes all his research time.
- Beliefs: “This stuff should be easier to get a handle on”
- Environment: Macbook for development, a Linux desktop for test deployment, VirtualBox for testing on different browsers.
- Interface Usage Tendencies:
- GUI: High
- CLI: Low
- API: Low
- Collaborates With:
Persona #4: Martina the Senior Sysadmin
Martina is a senior systems administrator at a large university. She has to manage more systems than she really has time for, and much of her time is interrupted by requests from grad students and other members of the university for resources. The IT department has some automated systems for deploying new sites with Drupal, Wordpress, MediaWiki or the like, but it needs to be done by hand. Martina's group has recently deployed an OpenStack environment, allowing some groups to provision virtual machines by themselves. This is some help, but is a level too low to take care of most users' real needs. She's interested in taking this one step further and providing a full Platform as a Service offering. By-line: "I could automate this, if I weren't so busy."
- Adoption curve: Early adopter
- Skills: Shell scripting expert, hacks at Python, Perl, and just about anything else. Good at putting all the parts together.
- Behavior patterns: Would like to work on interesting emerging technologies that she believes will really help her employer and her users, but keeps getting interrupted by break/fix and user help requests.
- Goals: Provide a self-service application infrastructure for users.
- Needs: Robust deployment tools, simple automation stack, ability to rapidly provision large number of hetergenous systems at once, and then efficiently monitor and change them.
- Main Tasks: Provision new systems on both the public cloud and the private OpenStack cloud her company runs. Work with developers to deploy code as often as possible.
- Attitudes: -
- Beliefs: “Whenever a sysadmin has to interact with a user directly, something has gone wrong”
- Environment: Fedora on her desktop, Android on a phone
- Interface Usage Tendencies:
- UI: Low
- CLI: High
- API: Medium-Low
- Collaborates With: Developers, junior administrators, CTO, finance organization
Persona #5: Sarah the Data Scientist
Sarah is a data scientist in a biotech startup. She works with large data sets and intends to process them with map-reduce. She plans to test this in her office and then scale out to local cloud resources and to spot instances in Amazon EC2 By-line: "Crunching a large set of data, get it done and done efficiently"
- Adoption curve: innovator
- Skills: Python, Scala, Octave, R, Hadoop, Cassandra, and Pig.
- Behavior patterns: Most interested in results, but not afraid of tackling new technology to get there.
- Goals: Process data quickly and efficiently, at the lowest cost possible. Since Sarah is potentially working with medical data, she cares about data strorage, retention policies, and ethical issues around her data - not just reliability and cost.
- Needs: -Reliable images with the latest improvements in her toolchain. Does not want to compile her own Hadoop, Pig, Mahout, etc.
- Main Tasks: collecting/crunching data, creating visualizations from data, managing data storage/retention.
- Attitudes: she cares about performance and is energy-conscious. Cares about security and the ethics of data management.
- Beliefs: "The tools are only a means to discover interesting scientific and user data"
- Environment: Fedora on her workstations, CentOS servers in the data center (including an OpenStack deployment), uses public clouds like Rackspace, the HP cloud, or AWS spot instances to spin up many servers at once for a short period of time, a Mac laptop
- Interface Usage Tendencies:
- GUI: Medium
- CLI: Medium
- API: Medium
- Collaborates With: Other researchs and scientists, lab technicians, HPC techs, more 'pure' developers
Persona #6: Jody the HPC Scientist
Jody is a researcher at a major university. She previously used grids to solve meteorological simulations using batch schedulers (SGE, Torque). Since she needs to run many parallels simulations, she thinks about using cloud ressources for temporary infrastructure overflow. By-line: "I need a bunch of machines to run my computational tasks, cloud or not."
- Adoption curve: Early majority
- Skills: Shell, Python, Fortran
- Behavior: Most interested in results, but not afraid of tackling new technology to get there.
- Goals: Run her computational tasks
- Needs: access to distributed filesystems (NFS, GlusterFS)
- Main Tasks: Designing experiments and running them
- Attitudes: She is not a computer nerd, but someone who likes to get things done
- Beliefs: "The tools are only a means to validate my theories and give them real-life applications"
- Environment: Fedora on her lab workstations, CentOS servers in the university data center (including an OpenStack deployment), uses public clouds like Rackspace, the HP cloud, or AWS spot instances to spin up many servers at once for a short period of time, a Mac laptop
- Interface Usage Tendencies:
- GUI: Medium
- CLI: Medium
- API: Medium
- Collaborates With: other scientists, few research engineers (sysadmin or developers)
Primary Use Cases
Web Application Deployment in a Public Cloud
Modern web applications are deployed as a collection of interconnected services, including parts like web servers, application servers, databases, and caching layers. Fault tolerance is handled at the overall orchestration level rather than by individual instances Fedora Cloud can be the base of each of these parts, providing recent libraries, server software, and language toolchains. Each system will be managed using the public cloud's own management tools plus a configuration management system like Puppet, Chef, Salt, or Ansible.
Web Application Deployment in Private Cloud
As above, but in a locally-deployed and managed private cloud system.
Web Application Deployment in Hybrid Cloud
As above, but rather than a single cloud provider, the application seamlessly takes advantage of resources in both a private and public cloud.
Big Data / Number-Crunching
Deploy scale-out application for data processing to public, private, or hybrid cloud.
Docker Container Host
Docker is a technology for running applications inside a protected container. These containers run under the host kernel, but otherwise are self-contained. The Fedora Cloud image running in either a private or public cloud can provide the host level, including basic docker management tools plus tools for security and for access to storage. Workflow may involve developing and testing Docker containers on local system and pushing to cloud for production.
OpenShift Origin System
OpenShift Origin is an open source platform-as-a-service system already included in Fedora. A Fedora Cloud image focused on OpenShift would make it easy for users to run their own PaaS.
Simple Deployment of code to from dev to production
As a developer, I want to ensure that my code is easily deployable from my development environment to a production environment, without encountering issues of non-compatibility. Example: A feature described as “Improved deployment of code to a cloud” might be fully described in the Features section, and as a result, Vagrant might be listed as a requirement in the non-functional requirements section, and cross-coordinated with the workstation working group. There are likely numerous other requirements as well.
Development as part of a team
As a member of a development team, I would like to develop in an environment where code can go through unit or functional testing, and be approved or accepted by other members of the team. Example: A feature described as “Continuous Integration platform” may be listed in the Features section, and the various tools available to implement would be enumerated and described in the “non-functional requirements” section, and cross-coordinated with the server working group.
Development using libraries not included in the distribution
As a developer, my toolchain includes libraries or dependencies on other packages using libraries not in the Fedora distribution, and I may be developing for distributions not limited to Fedora. Example: A feature described as “Portable code” (REALLY POOR DESCRIPTION, sorry) may be listed in the Features section, and “Software Collections” might be listed as a requirement. Note that this may have some cross-over with a possible additional story/use case, “Deployment of code using libraries not included in the distribution.”
Target IaaS environments
The Fedora Cloud product can be used as a guest/VM/image under many IaaS services and providers. For projects that are not currently packaged within Fedora/EPEL, we may need to locate a kind person to ensure testing. These include: Open source IaaS systems:
- Apache Cloudstack
- Amazon EC2
- Google Compute Engine
- HP Cloud
- Digital Ocean
Features here address the primary and secondary use cases, product or secondary objectives, market opportunities from above. Features should provide functional requirements (“what it should do”) preferably in a narrative fashion - more of a story / solution description, rather than “package XYZ” - the features (the ways to meet a user's objectives?) each likely consist of more than one package/enhancement, and those packages should be detailed in the “Detailed requirements” section of this document, and each of those detailed requirements should refer back to which feature it supports.
Feature description should be described in the line saying “Feature #1/2/etc.”. Describe the feature in more detail, specifically addressing how it addresses user scenarios, primary or secondary use cases / objectives of the product. Use a table to indicate the following items: Priority (Must, Should, NTH) Citation of use cases addressed As work continues and specific detailed requirements are developed, reference the detailed requirements within this document helping to fulfill this feature. This helps to ensure awareness around “do we still have a feature if some of the detailed requirements are not fulfilled, and thus are not able to address the specific use case needs / user objectives.”
Feature: Cloud Bootstrap Tools
The Fedora Cloud guest image includes cloud-init, the defacto standard for boot-time configuration for cloud instances. It is a critical tool in getting public and private cloud instances properly configured based on metadata present to the hypervisor, like the desired hostname and filesystem configuration of the guest. Heat, one of the OpenStack sub-projects is also playing an important role in the orchestration landscape. We include the client tools for Heat bootstrapping. The cloud landscape is rapidly changing and we will consider additional tools to help with guest initialization and setup as they appear.
Feature: Ready Access to Fedora Collection of Packages
language and application stacks to enable whatever you need... blah blah blah FIXME
Feature: Docker support
Docker is an easy to use interface for running application containers on Linux. Fedora is uniquely positioned to provide the best platform for Docker, since this container technology is not a security solution, but can be made reasonably secure when wrapped with SELinux. This includes adding libvirt support to the image, which is more heavyweight than many users of the image for other purposes will want, so we will produce a dedicated image specfically tailored for Docker.
Feature: Big Data Tools
... FIXME. .... Exact contents in collaboration with Big Data SIG ....
Feature: Cloud -> Server
Easy path from base cloud image to Fedora Server
Feature: OS Tree
??? FOR DISCUSSION ??? The Fedora Cloud image includes OSTree for safe, atomic updates.....
TODO: change these things into actual requirements in the lovely format shown below.
- Support for smallification: a) kernel (reqs for kernel team), b) lang & docs (reqs for packaging tools team), c) cloud-init refactoring
- Docker security ( = strong selling point): Libvirt/SELinux integration for Docker (planned but not done)
- Infrastructure for automatic production and upload of cloud images for updates
- easy software stacks -- install your language of choice, preferably your version of things of choice
- integration with Fedora Server roles
Requirement #1 (Short Description of Requirement)
Refer to which previously described Features, Use Cases this requirement helps to fulfill.
Must, should, NTH
High, Medium, Low
Stakeholders / Owners
Any major dependencies, including things that may require any cross-working-group coordination, should be called out here. Any process changes required within Fedora should be documented here as well.
Level of testing required; is it a blocker to release? Is the testing automate-able?
- Existing BZ:
- Upstream webpage / wiki page / github page(s):
Requirement #2 (Short Description of Requirement)
These are the requirements needed that are not necessarily part of implementation of the product itself, but are still required as part of either making the product more attractive/useful, compatibility requirements for a user's workflow (ie: “works with Puppet,”) or things needing to be done/coordinated in other areas of the project (in other working groups?) to ensure a well-rounded solution. For each, as was done with Features in a previous section, we should call out some additional information, assuming doing so makes sense for the requirement:
- Priority (Must, Should, NTH)
- Effort required
- Major dependencies / Process Changes needed
- Existing Documentation: BZ #, pointers to upstream project docs
How to create images. Do not be shocked if there are 48 ways to do this. Also: not sure if we want to include containers in this section.
Creation of “official” Fedora Project images
New images. Updated images (ie: providing newer images for a release that have updates already included) might also be included here?
Images created by users
Migration / Upgrades
Performance / Scalability / Failover
Fedora Project Documentation
Open Source Projects documentation
Ensuring that Fedora is well-represented, up-to-date in other open source project documentation...
- at some point: ARM (which variants ?)
Virtualization types? Container types?
Internationalization / Localization
- NTH: Cloud Product base images should only includes en.us_US locale since it is meant to be used as an deployment target to save space. Other locales should be available through repositories.
The idea is to rely on langpacks rather than RPM wizardry.
Logging / Auditing
Monitoring / Notification
About this Document
This PRD (Product Requirements Document) is an evolving document, created by the Fedora Cloud SIG Working Group as part of the process for designing the Fedora Cloud product. The framework for the PRD itself is currently in a draft state. This document will evolve over time as the purpose of the SIG continues to be determined as the working group process gets under way and initial products start to get produced.
Contributors to this document include:
Reviewers & Contributors
The following people have contributed to the development of this document, through feedback on IRC, mailing lists, and other points of contact.
The Fedora Cloud SIG is one of many teams within the Fedora Project. The Cloud SIG mailing list is located here. Minutes and logs from IRC meetings related to the development of this document should be listed here as the document evolves.
Over time, it is expected that this document will undergo various rounds of review, approval, and editing; in the future, it may be rewritten for different releases of Fedora. While one can review the history of a wiki document (by clicking the "history" tab), it is useful to provide explicit indicators of any major format changes, approvals, or indications of it being in a “final” state, in a list that can allow someone to quickly see that all of the prescribed layers of approval have occured.
- January 8, 2014: Collaborative hackfest day.
- October 28, 2013: Initial Draft of template.
- FutureDate: Approval by SomeGroup; link to any pertinent mail announcement and/or meeting minutes
Tracking of Progress
This document contains numerous descriptions of use cases, descriptions of feature sets to address the use cases, and the requirements to enable those features. Numerous Fedora self-contained and systemwide changess (in addition to updates to individual packages) may combine to address those use cases and feature sets. Thus, as a single release, or even series of releases, undergoes development, it is useful to easily track how an entire use case or feature set may be progressing. While Fedora uses the Changes Process to track changes in the distribution, those changes are typically described as details of changes to a specific package, or the introduction of a specific package, rather than as a piece of a larger feature set. This document could possibly be used to do any or a number of the following things:
- Provide a secondary location where changes are tracked (which seems like a lot of overhead to me)
- Provide a location where overall Feature Progress is tracked, via periodic cross-checking against Change pages; this could be either in a standalone section, or simply attached to each Feature description.
- Scope out how features are expected to progress over a number of releases.
- None of these things.
When we more fully determine how to most efficiently track progress, the pointer to where that tracking is done, and/or the description of or process by which we do the tracking is formalized, should be documented in this section in lieu of what is currently written here.
Definitions and Acronyms
- AWS: Amazon Web Services
- Amazon EC2: Amazon Elastic Compute Cloud, a popular public IaaS
- IaaS: Infrastructure as a Service
- PaaS: Platform as a Service
- SaaS: Software as a Service
- PRD: Product Requirements Document
- EPEL: Extra Packages for Enterprise Linux
- CI: Continuous Integration
- CD: Continuous Delivery or Continuous Deployment
- SCL: Software Collections
- teams in charge of some aspects of Fedora Project
- NTH: Nice-to-have
- BZ: Bugzilla
- GUI: Graphical User Interface
- CLI: Command Line Interface
- API: Application Programming Interface
Indication of prioritization
- Items marked with a ? and following question in italicized lettering are open questions.