Another impressive OpenStack Summit has now concluded. The community works hard for months, excitement builds up for weeks and then, in what seems like a blink of the eye another OpenStack Summit is in our rear view mirror. As with the past conferences, Vancouver delivered over and above all expectations. Even the weather gods seemed happy - five continuous days of bright sunshine over Vancouver downtown - I say it had something to do with the collective positive karma that the OpenStack community brought along!
From a networking (Neutron) perspective, significant progress was made, with more planned and discussed during the Liberty Summit design sessions (check out the article below). Like with all maturing engineering projects, it was obvious that Neutron community will continue solidifying the core (simple, reliable networking, operationally resilient, consistent APIs) and at the same time extend the innovation frontier (container networking, federated pods). Providing a consistent way to deal with physical, virtual and container(ized) workloads and associated network connectivity and policies, while eliminating issues such as traffic hair-pinning that have been concerns in the past will be critical as we move forward. At Big Switch, we will play our part in this effort and will continue to be an active part of the community.
With all the activity, this past week was certainly exciting but also a bit overwhelming. So many interesting sessions and discussions going on in parallel, there is simply no way to stay on top of it all. Thankfully, almost all of the sessions are recorded and available here: https://www.openstack.org/summit/vancouver-2015/summit-videos/.
As we sign off from the OpenStack Vancouver Summit, I want to again thank all our customers, community and ecosystem partners. We could not have done it without your help. See you all in Tokyo!
-- Sunit Chauhan
Notes from Day 4:
In the last two Neutron design summits, the emphasis was on paying down the technical debt and achieve feature parity with Nova networking. Few new features were introduced. In the Liberty Summit, the community regained its stride on innovation and proposed a slew of new feature proposals. There were a few that made me excited:
- Address Scope and Subnet Pool. The proposal was led by Carl Baldwin. The idea is to give an operator a way to manage tenant subnet allocation. This feature will enable quite a few new use cases for Openstack Neutron. IPv6 address allocation was the primary motivation behind the proposal. It will also enable IPv4 deployment to remove NAT bottleneck from its north-south traffic. The idea is to enable a tenant to request a subnet from a routable subnet pool, instead of the current work flow of assigning a private subnet on its own. When a tenant network receives a routable subnet from the pool, all workloads from the network can directly reach or be reached from external networks without NAT. It will be a very useful use case for Enterprise private cloud, where the cloud network can be on a subnet that is routable within the corporate network without NAT. The proposal will also benefit the next feature, BGP.
Etherpad link: https://etherpad.openstack.org/p/YVR-neutron-l3
- Dynamic Routing APIs: Dynamic routing has been one of the top asks from customers deploying OpenStack in their data centers. A provider workflow to enable prefix exchange and ECMP with the upstream routers will remove quite a few friction points in Openstack enterprise deployment. The proposal provides an API abstraction for BGP peering configuration and ability to define routing policy in the near future to influence the route based on application requirements. Address scope will simplify the prefix aggregation for route advertisement.
Etherpad link: https://etherpad.openstack.org/p/YVR-neutron-l3
- Pluggable IPAM is another great feature that will land in Liberty. Neutron can finally get out of IP address management and allow external entity to provide the management. It enables integration with existing IPAM management system for enterprises.
The collection of the above 3 features will make Neutron integration with enterprise network a much friendlier exercise. Additionally, progress was made on the Services front as well:
- Firewall as a Service: FWaaS will finally achieve its rightful place after being experimental features for the last four releases. It is great to see the community come to a consensus that FWaaS addresses a much broader security need than Security Groups could. Keeping the two in separate APIs will encourage innovation and feature velocity for both. In the meantime, the community emphasized the need to identify a common object model for both APIs to avoid end-user confusion. Both developments were well received and made great progress. Hopefully, FWaaS will be part of Neutron service API soon. Then, we can move on to tackling the service insertion APIs.
- Tap- as a Service: There were quite a few proposals and demos on enhancing the Neutron operations as well. Tap as a Service (TaaS) stands out in providing the operator a quick and easy way to monitor traffic of interest. It defines both tenant and provider workflows to tap any neutron port and forward the tapped traffic to another neutron port, where a monitoring VM connects. The proposal enables on-demand monitoring and opens doors to wide variety of traffic analytics and troubleshooting tools.
On the whole, it was a great Summit for Big Switch’s engineering team. Many of the features that we have been working on for sometime are now going mainstream with well-defined APIs. Its exciting to see the Neutron mature and at the same time continue pushing on the innovation frontier!
-- Kanzhe Jiang (MTS at Big Switch)
Notes from Day 3:
As we mentioned in an earlier post, OpenStack partner ecosystems are maturing and enterprise data center architects and deployment teams are looking for end-to-end, production-grade solutions. In our conversations, it was obvious that they are now demanding mature solutions to build out their OpenStack private clouds – compute, storage and networking infrastructure, combined with best-in-class tools and services.
At Big Switch, we play our part by not only delivering the best Leaf-Spine Clos Networking Fabric for an OpenStack data center but also by seamlessly integrating solutions from a range of technology alliance partners. These partners span across different layers of the stack - orchestration, L4-L7 services and hardware platforms. By providing choice at each layer, the OpenStack community is finally providing the desired freedom from lock-in solutions. We are very excited to be part of this open ecosystem.
For Big Switch, this mutual respect and belief in a common, open future was fully evident at the Vancouver Summit – both on the Expo floor as well as our partner presentations and sessions. Below are some highlights of activities with each of our key partners at the conference:
- Mirantis: Big Switch was showcased as a “Mirantis Solutions Partner” at their booth. Our joint data center network resilience study – Chaos Monkey white paper – was a very popular discussion topic with many attendees looking for production grade validation. Finally, thanks to Nick Chase of OpenStack:Now newsletter fame, for conducting featured interviews. You can check out the videos here:
- Dell: In addition to Big Cloud Fabric solution being prominently highlighted in the Dell Expo Floor booth, Prashanth Ganapathy from Dell, talked about our joint solution in his presentation on the modern, open networking. Check out the Dell session video here.
- Additionally, we had great collaboration with our partners Red Hat, Canonical, A10, Nexenta and F5 among others – which in some cases even involved real-time code development at the conference – talk about continuous collaboration!
As we get close to wrapping up the Vancouver Summit, we would like to express our sincere thanks to all our alliance partners for the joint outreach efforts. We appreciate the trust you place in our technology and solutions and look forward to even bigger things in the near future.
-- Bala Ramachandran & Sunit Chauhan
Notes from Day 2:
OpenStack Design Summit: Packaging Work Session
On Day 3 of the OpenStack summit (Room 218 - May 20th, 4:30 pm) I will be helping moderate the Ops track session on packaging: Packaging Work Session. An outline for the session which is available here.
When we discuss packaging in Openstack what we are really discussing is how we set up code to be moved around and made part of several work flows in the production life cycle of an OpenStack environment. From integration to CI systems, through the supply chain of the software stack, all the ways to the actual deployment of code onto compute resources whether physical or virtual.
OpenStack is written in python, and due to the complex nature of the project ecosystem has run into some sizable issues internal to the project focused on maintaining a safe dependency tree. As deployers look to make OpenStack integrate well with their own existing environments and services they find themselves needing capabilities that currently do not exist with traditional RPM or debian packages. Such as the ability to install multiple versions of the same python module, or satisfy a compatibility matrix that has never been tested by the OpenStack community.
One interesting development that we will be discussing is the introduction of the Ops tag. Which you can read more about here. Today's discussion should be a good one, and a healthy view into one of the areas of OpenStack that still has some rough edges in need of smoothing.
Hope to see you in room 218 at 4:30 pm!
-- Matt Joyce
As expected, the second day of the OpenStack Summit Expo got off to a busy start for the Big Switch team. A few of us attended the Keynote Session led by Mark Collier (OpenStack COO) and it was great to hear him share his thoughts on the evolution of this movement. Check out the attached video link on how OpenStack has moved from experimental stage to maturity and now to broad adoption stage.
The team also had the opportunity to speak with a range of fellow OpenStackers – at our booth on the Expo floor as well at the Networking sessions. As mentioned in our update yesterday, the maturity of the technology is quite apparent. From our conversations, one can clearly see two categories of users appearing – the early adopters followed by the early majority:
- Early Adopters: This includes super users we have seen over the last few Summits – good examples are companies such as Google, Walmart Labs, Yahoo etc. These companies typically have deployments of 100s of racks in their data centers. They have been using OpenStack for years now. They continue to lead the cutting edge innovation and are now very clearly involved and leading next generation of OpenStack efforts around Containers (Project Magnum) and App Ecosystems (Project Murano).
- Early Majority (aka the Enterprise): These include teams of infrastructure architects and engineers who have been part of the Open community, contributing and participating over the last few Summits but are now comfortable with the quality and reliability of the solutions the industry is making available to them. They are looking for proven, tested and supported OpenStack solutions – fit-for-purpose and ready-to-deploy!
From an ecosystem perspective, Mirantis, the Pure Play OpenStack company, has been at the fore front OpenStack solutions and services. It is no different this year. Red Hat and Canonical, our partners are participating in full force as well. In addition, this year we also seeing VMware at the summit. In their sessions yesterday, they discussed VMware Integrated OpenStack (VIO) solution (and integration through vCenter/NSX/vRealize Suite) with some of their existing customers such as Adobe, Nike and Wells Fargo.
Its an exciting time to be part of the OpenStack movement!
-- Sunit Chauhan and JP Shukla
Notes from Day 1:
I have attended the OpenStack Summits over the last couple of years and I am always impressed with the progress made in such short 6 month sprints. This year is no different – there is excitement all around in Vancouver and everything is just that much bigger in scale!
The Big Switch team got off to a busy start – sessions to be attended and presented at, scheduled (and more importantly unscheduled) meetings with fellow OpenStackers, and getting everything setup for the rest of the week. Thanks to JP, Matt and Bala for rolling up their sleeves and we got an awesome booth on the Expo floor – could not have done it without their help! Later in the day Rob Sherwood, our CTO, was on the Startup Founders panel and had a great on-stage discussion on the challenges and future of OpenStack networking. And it all concluded with a very busy evening on the Expo floor.
The key theme I have seen emerging this year is that OpenStack is now a mature technology. The discussions about the scale and stability of early deployments and the need for validations from a few select super users has given way to in-depth conversations about real life deployment experiences – at enterprises of all sizes and scales. Yes, there is still a lot that needs to be done; there’s always the next frontier (containers anyone?), but OpenStack super users are all around here in Vancouver. We heard from a bunch of them on the first day are more scheduled today.
From a networking perspective, my discussions yesterday centered around a couple of themes:
- Chaos Monkey Grade Infrastructure: Ok, I’ll admit that I just coined that term, but the need for tested, resilient solutions that one could take into production quickly is very real. OpenStack solutions that just work – easy to deploy, simple to manage and resilient to failures.
- Ecosystems are Emerging: Mainstream companies are looking for complete packages and the industry is maturing to satisfy those needs. A big thank you to our partners at the Summit – Mirantis, Dell, Canonical, RedHat, A10, F5!
Looking forward to all the action on Day 2!
-- Sunit Chauhan
Meet Big Switch Networks @ OpenStack Summit Vancouver
With less than a week to go before the start of the OpenStack Summit, excitement is in the air! At Big Switch, we are looking forward to meeting fellow OpenStackers and share the latest learnings from production deployments of OpenStack Pods.
Join Big Switch to learn how you can benefit from simple, resilient and economical networking for your OpenStack data center:
- Visit Us in Booth #T73: You will not only get to see OpenStack Networking in action, but also get entered into a raffle for Apple Watch!
- Join Rob Sherwood, Big Switch CTO, as he discusses the Future of OpenStack Networking with his peers and 451’s Eric Hanselman (May 18th at 5:30pm)
- Experience Chaos Monkey Resilient Fabric: Register here to setup time with our OpenStack & Big Cloud Fabric product team in Vancouver