Dave Neary is part the Open Source Program Office at Red Hat, driving adoption and community growth for projects including oVirt, RDO, a distribution of OpenStack for Red Hat-based operating systems, Gluster, and Fedora. Dave is currently concentrating on projects related to developer tooling, in particular the awesome Eclipse Che. He previously worked for several years on Network Function Virtualization and Software Defined Networking. Dave has been active in free and open source communities for more than 15 years as a consultant, community manager, trainer and developer. Find him on Twitter @nearyd
In the previous article, we outlined how cultural dissonance can cause issues when cultures collide. In this article, we talk about what makes up cultural identity, and perhaps help you become more aware of your own cultural assumptions.
In "Cultures and Organizations", the sociologist Geert Hofstede identified six dimensions for characterizing a culture. The effects of these dimensions can be analyzed according to multiple characteristics of a culture, including education, social structure, political and economic systems, religious attitudes, traditions, and customs.
Communities are messy. What makes them messy is the relationships between the humans who make up those communities. Humans are complex beings, each with their own sets of experiences and assumptions which they bring into a relationship, and when the underlying assumptions about how the world works do not match, we get into trouble.
One colleague of mine has described culture as "the way that humans in a community actually do things." The things we say and do consciously are the tip of the iceberg—many of our reactions and feelings are unconscious, and based on underlying values and assumptions about society which we may not even be aware we have.
Editor’s Note:Speaker notes for FOSDEM 2017 talk in Community DevRoom (video included).
For anyone reading this who has been a parent, you will know that the most surprising thing about toddlers is how fast they change. You go away for a couple of weeks of business travel, and when you get back, there’s a different person waiting for you when you get home. And two years old is about the time when the rate of change is at its peak.
That is what OPNFV feels like now: we have navigated the early teething issues and reached a state where there is lots of change and lots of progress. We have labs on every continent, many active projects developing new NFV related functionality, three platform releases, and increasing industry interest. Two and a half years ago, NFV was a promising architecture concept. Today, operators are putting NFVi platforms into production, thanks in part to the work of OPNFV.
I have been thinking about how size can affect culture and adaptability of groups recently. The topic once again came up today in a talk about what makes a healthy community. The answer to that will depend on the community’s size and maturity. An open source project, in the words of one participant in one conversation I had recently on this subject, should have "the minimum level of structure to allow it to function effectively." I agree—just enough is the right amount. This article contains some ponderings on the relationship between size and communities, and some conclusions we can take from that.
In August 2015, George Zhao of Huawei, formerly the OpenDaylight release manager, was assigned to be OpenDaylight community manager full time, a role that I had been filling on a part-time basis since October 2014. To help him ramp up as a first-time community manager, I agreed to mentor George. In the course of working together, I have had the opportunity to structure some of the things I have learned in my career, and pass them on to him.
This series of articles, resulting from my conversations with George, is a collection of personal thoughts and analysis on community management, which I hope will be useful to others.
NFV (Network Function Virtualization) has taken the telecommunications world by storm in recent years. Communications service providers plan to run their core services in virtual machines, running on a standardized, open source platform. This promises to reduce both capital and operation costs, and most importantly, to accelerate the delivery of new services to market through increased agility.
The Open Platform for NFV (OPNFV) project is deploying a collection of open source projects, including OpenStack as the Infrastructure-as-a-Service layer, OpenDaylight and Open vSwitch for virtual network, Ceph for virtual storage, and libvirt and qemu/KVM for virtual compute. OpenStack and OpenDaylight are themselves collections of multiple projects, each with their own set of maintainers. All told, the OPNFV platform consists of around 20 different projects, many of which will need to be changed to satisfy the performance and reliability constraints of an NFV platform.
How vendors make and distribute those changes will vary from one to another. Some participants will attempt to maintain differentiation by developing features in private, and integrating them in their products without ensuring the changes will be integrated into the upstream open source projects. Others will propose changes at the same time as they are released in a product.
Red Hat is a company with a policy we call “upstream first” – we work to get features integrated into open source projects before we integrate them into our product offerings. Wouldn’t it be easier just to develop features for our customers, and let the upstream projects figure out what they care about? What are the costs associated with the upstream first approach? Why do we take this approach?
The short answer is because it is the cheapest, most sustainable way to innovate on an open source platform. To explain why, I will explore what it means to build on top of open source projects, the different approaches people take to doing it, and the costs associated with each approach.
For the first time, FOSDEM will include a dedicated track for networking. The DevRoom covers two distinct topics: network management (the configuration of host-level networking, such as VPNS, and ethernet and wifi connections) and SDN (Software Defined Networking, the management of networking in large networks, including routing and switching traffic, and enforcing network policy).
Software Defined Networking
The DevRoom opens with presentations on Software Defined Networking, the ability to define network flow centrally for an entire network. Presentations cover three themes:
SDN controllers, overlay networks: These projects can manage networking for clusters of hosts, potentially with many virtual machines running on them. An SDN controller can manage both physical and virtual infrastructure centrally, whereas an overlay network completely dissociates the virtual network topology from the physical network topology. Learn how to use OpenContrail as a network virtualization platform with OpenStack, or discover newer open source projects including Calico, MidoNet, and VXVDE. We will also have an overview of OpenLISP, an open source implementation of the Locator/ID Separation Protocol.
Virtual switches: A virtual switch is a way to bridge network connections for multiple virtual machines on a host to the physical network. Thomas Graf will give us a guide to tracking stateful services with Open vSwitch.
Packet filtering, dataplane processing: A key piece of virtual networking is the ability to move network packet processing functionality out of the kernel and into userspace, where it can be less of a bottleneck. There are a number of presentations on the high-performance packet filtering project pflua, and on the Dataplane Development Kit (DPDK).
The second part of the day focuses on network configuration inside a host — automating the process of connecting to the network and managing network connections and network hardware.
Hardware management: Learn more about OpenNMS, open source solutions for managing hardware switches, and using the netlink protocol with pyroute2.
Network management projects: There will be status updates from NetworkManager and networkd.
Network services: Learn about Knot DNS, a stand-alone DNS server.
For this first edition, we will have 16 half-hour talks, which allows attendees to get a brief overview of a lot of projects. All of the presentations will be recorded and will be online soon after the conference.
For more information about the projects presented, check out these links:
Throughout the conference, finding two, three, or four networking or SDN sessions occurring at the same time was not unusual. In the main conference, there were 15 sessions specifically on NFV and related topics, many to standing-room-only audiences, and another 30 or so on networking and SDN. In the developer conference across the street, the OpenStack Kilo Design Summit, two sessions were dedicated entirely to NFV and telco needs. In addition, it was a prominent theme all week in design summit sessions related to OpenStack projects Nova (compute), Neutron (network) and Congress (policy).
Whether you are deploying applications in containers, VMs, or on bare metal, being able to group related services together into private networks and control the traffic flowing into and out of your infrastructure is important. The SDN controller fulfills that role by allowing you to define network policy centrally and have that policy applied at the edges of your network, in the physical and virtual network switches.
OpenDaylight is an open source SDN controller. In its short lifetime, OpenDaylight has gained support from a diverse set of companies and individuals who are eager to see an open source controller serve the networking needs of traditional IT, cloud infrastructure platforms, traditional virtualization management, and fleets of containers. Cisco released the initial code in 2013 and the project now includes 41 paying members. Red Hat is a founding member of the project at Platinum level.
This week the OpenDaylight project had its second major code release, code-named Helium, which is a big advancement for the project. The release includes more than 4,700 contributions from 183 engineers, representing 20 companies. More than 300 commits came from the affiliation "independent", which also shows the breadth of the project’s appeal.