How to Avoid the Cloud Trap

Do you ever want something so badly that you might just fall into a trap trying to get it? That cheese looks awfully yum…whack! Oops :(

When speed = haste 

When speed = haste we blind ourselves to potential pitfalls. One area today of much haste in many enterprise IT organizations is, you guessed it, the move to the cloud. I was visiting a customer today, and I was told a familiar story: Their application team has built a new application proof-of-concept on AWS, and the line of business is going to fund this app to move ahead. The familiar part of the story line is that this was all achieved in X weeks instead of X months. With impressive results, developers are soon pulling the rest of IT to do cloud with an executive push.

Embracing cloud native is a great thing, but doing so with this haste is running right into the palm of AWS’s hand. Not to pick on AWS, this can happen anywhere, but it’s most likely to happen at AWS because they’re the incumbent choice, and there are so many services to help developers deliver something cloud-native quickly.

To developers, these AWS services look like candy to Garfield on halloween. Unsuspectingly, many will never see the trap. Why? Because the lock-in police within the IT organization are patrolling I&O, while the lock-in felons have cleverly gone to work over on the application developers.

“The new ‘lock-in’ is from developers,
not infrastructure and operations”

I often talk about the move to cloud being led by developers and the devops trend; however, I hadn’t put 2 & 2 together quite like I did today with respect to lock-in. Although, I do often preach about how 80-odd percent of enterprises are targeting the hybrid cloud model, and how to make hybrid cloud a true IT platform, there must be portability across clouds, I guess I didn’t see the inverse of that concern.

That brings us to discussing the solution. How to achieve portability and a hybrid cloud IT platform with, less but not absolutely no, lock-in.

Let’s just examine the main public cloud business models for their services.

1. First, nearly everyone is very familiar with IaaS. It’s offered at almost all big clouds, certainly the main 3.
2. Then, there is the managed service model, where the cloud provider will usually take the many popular open-source software projects and offer them as individual managed services.
3. Finally, there are the services that are homegrown or customized by the provider and offered as a service.

This is a decent summary of the main service models, but certainly not the only things that cloud providers can offer. Some other interesting examples include: AWS offers Direct Connect; GCP differentiates with its own high-speed global network to interconnect regions; Azure has the broadest compliance coverage. Anyway, if you understand the 3 models above, you can probably easily grasp what’s coming next… That developers that use customized, homegrown services are clearly locking themselves in the most.

Getting conscious about lock-in

So should you never use these customized cloud services? Of course you should if it is really worth it, but do it consciously. Obviously, services that are unique prevent apps that touch them from being very portable across your hybrid cloud platform.

The good news is that most cloud services today, have an open source project that matches most of what any service can do. If not, there is still a good chance that you can achieve similar benefit with a combination of open source projects, or vendor products based on such projects.

Bring your own stack

Perhaps the main thing that enables IT to go from multi cloud to consciously building a universal hybrid cloud platform is a unified toolchain and unified policy. Some systems’ policy unification can be achieved with meta-orchestrators that work across multiple clouds like Red Hat CloudForms and RightScale, but they have their limitations. Using the many benefits of public and private cloud IaaS, you can still control the application and devops stack to achieve portability and mitigate lock-in by bringing your own full stack that sits atop of any IaaS.

Anyone can do multi cloud. Throw in a little bit of this, and a little bit of that. But the recipe for a happier hybrid cloud seems to be known to some organizations, but not to all, so let’s have a look at how to do cloud with less lock-in and more portability.

Embrace cloud IaaS as a base for your devops stack, but use other cloud services sparingly:

• Bring your own IaaS automation and abstraction (such as Terraform and config management tools like Puppet, Chef, Ansible, etc.)
• Lock in to cloud services consciously when they are unique and necessary for business advantage
• For services that have open source tool equivalents, bring your own tool or at least use a managed service that has the generic API

Start with 2 clouds instead of one. This will…

• Prevent you from tethering yourself to just one partner for cloud innovation & economics
• Force the application cluster / stack to be portable
• Force the DevOps workflows to be portable
• Force designing for resiliency and scale early on

Take the naïve out of cloud native

If your business case for cloud is that your developers’ new-found speed is knocking the socks off of your executives, then maybe have a closer look.

There is a better way to “do cloud” with portable apps, automation, and software-defined infrastructure atop of any IaaS, and even better, there is actually plenty of help. Check out the cloud-native computing foundation that Juniper and many other vendors recently joined and for your unified toolchain check out some of Juniper’s cloud software portfolio that can fit equally well across your public/private/ hybrid cloud venues such as AppFormix, vSRX, vMX and Contrail Networking built from OpenContrail.

Juniper & Red Hat Serve Up an Open Double-Stack Cloud with an SDN Twist

Photo credit Tom Haverford

Juniper Networks Contrail Networking, developed in the OpenContrail open source project, has long been a part of Red Hat’s millinery. The partnership between Juniper and Red Hat goes back some years now. Collaborating on OpenStack cloud and NFV infrastructure has won these partners success in supporting large enterprises and communications service providers like Orange Business Services.

At the long list of open source festivities in Boston over the next 2 weeks, you will hear these partners in cloud building on their past successful OpenStack + Contrail integration and now putting the spotlight on new integrations to support cloud native. You’ve heard me blog about the OpenContrail integration with OpenShift back a year already (in its first alpha form that I demoed), and more recently for CloudNativeCon and DockerCon talking about how we evolved that work to make this integration enterprise-ready and up-to-date with all the innovation that’s happened in the fast-paced OpenShift releases.

But how do you get the best of OpenStack and OpenShift?

Red Hat has been helping customers to move faster with devops, continuous delivery, and containers using OpenShift for a long time. Naturally Red Hat often does this atop of Red Hat OpenStack Platform, where OpenStack creates clusters of virtual machine hosts for the Red Hat OpenShift Container Platform cluster.

One latent hitch in this double stack is the software-defined networking (SDN). Like OpenStack, OpenShift and Kubernetes (on which OpenShift is based) have their opportunities for improvement. One area that is frequently fortified and improved is the software-defined networking (SDN), and the importance of doing this doubles when you’re running the double stack of OpenShift on top of OpenStack.

Why can SDN be such a snag? Well, the network is a critical part of any cloud, and especially cloud-native, infrastructure because of the enormous volume of microservice-generated east-west traffic, along with load balancing, multi-tenancy, and security requirements; that’s just for starters. The good-enough SDN that is included but swappable in such open source cloud stacks is very often indeed good enough for small cloud setups, but it is common to see the SDN replaced with something more robust for clouds with more than 100 nodes or other advanced use cases like NFV.

Fast and Furious Clouds

I think of this 100-node edge like the 100mph edge of a car… As I make my way to Boston, I just took an Uber to SFO, and of course it was a Prius. Like most cars nowadays they’re very efficient, and great for A-to-B commuting. But also like most cars, when you approach or (hopefully don’t) pass 100mph mark, the thing feels like it is going to disintegrate! I was a little uneasy today flying down the 101 at just 80-something mph. Eeek!

Now I love to drive, and drive fast, but I don’t do it in a bloody Prius! My speed-seeking readers will probably know, as I do, that if you go fast in a sportscar, it is a way better experience. It’s smooth at high speeds, and the power feels awesome. Now let’s say you also aren’t just driving in a straight line, but you’re on the Laguna Seca Raceway. To handle cornering agility, gas and maintenance pit stops, and defense/offense versus the other drivers, you need more than just smooth handling. The requirements add up. I think you get my drift…

My point is that, similarly, if you’re going to build a cloud, you need to consider that you’re going to need a lot more than good enough. Good enough might do you fine for some dev/test or basic scenarios, but if you need performance, elastic scale, resilience, F1-pit-crew speed and ease of maintenance, and a security defense/offense, then you need to invest in building the best. Juniper has been helping its customers build the best networks for over 20 years. It’s what Juniper is known for: high-performance and innovation. When you (or say NSS) compare security solutions, again, Juniper is on top for performance, effectivity (stopping the bad stuff) and value I might add.

When Compounding Good-enough Isn’t Great

Imagine the challenges you can run into with good-enough networking… now imagine you stack two such solutions on top of each other. That’s what happens with OpenShift or Kubernetes on top of OpenStack.

In this kind of scenario, compounding the two stacks’ SDN, as you demand more from your cloud, you will double complexity and twice as quickly hit network disintegration!

Ludacris Mode for Your Cloud

If you want to drive your cloud fast and furious, and not crash, you need some racing readiness. OpenContrail is designed for this, and proven in some of the largest most-demanding clouds. No need to recount the awesomeness here. It’s already well documented. Before you green-light it though, there is one thing we’ve needed to iron out: How does an OpenContrail double stack drive?

SDN Inception

When OpenContrail developers were in their first throws of SDN integration for Kubernetes and OpenShift, we often ran it inside of an OpenStack cloud. And what was the OpenStack SDN? OpenContrail of course. Yep, that’s right we have OpenContrail providing an IP overlay on top of the physical network for the OpenStack VM connectivity, and then inside of that overlay and those VMs we installed OpenShift (or Kubernetes) with another OpenContrail overlay. It turns out this SDN inception works just fine. There’s nothing special to it. OpenContrail just requires an IP network, and the OpenStack-level OpenContrail fits the bill perfectly.

In fact, SDN inception is pretty common, but not usually with the same SDN at both levels. The main place this happens in practice is because we run cloud-native CaaS/PaaS stacks like Kubernetes, Mesos, OpenShift paired with OpenContrail on top of public clouds, and that public IaaS line AWS has its own underlying SDN. It provides the IP underlay that we need in those cases.

What about when we control the SDN at the IaaS AND the CaaS/PaaS layers? Even if 2 SDNs (the same or different solution) work well stacked atop each other, it’s not ideal because there is still double the complexity of managing them. If only there was a better way…

A New Hat Stack Trick

This is where the OpenContrail community was inspired to raise the bar, and the Red Hat stack of OpenShift on OpenStack is the perfect motivation. What’s now possible today is to unwind the SDN inception and use one single control and data plane for OpenShift or Kubernetes on top of OpenStack when you run OpenContrail. The way this is realized is by having the OpenStack layer work as usual, and using OpenContrail in a different way with OpenShift or Kubernetes. In that instance, the OpenContrail plugin for OpenShift/Kubernetes master will speak directly to the OpenContrail controller used at the OpenStack layer. To collapse the data plane, we have a CNI plugin passthru that will not require the OpenContrail vRouter to sit inside the host VM for each OpenShift/Kubernetes minion (compute) node. Instead the traffic will be channeled from the container to the underlying vRouter that is sitting on the OpenStack nova compute node. We’ll save further technicalities and performance boost analysis for an OpenContrail engineering blog another day.

Juniper and Red Hat work on this latest innovation of flattening the SDN stack is coming to fruition. It is available today in the OpenContrail community or Juniper Contrail Networking beta, and slated for Juniper’s next Contrail release. As to that, stay tuned. As to catching this in action, visit Juniper and Red Hat at the Red Hat Summit this week and the OpenStack Summit next week. We’ll see you there, and I hope you hear about this and more OpenContrail community innovations ahead and in deployment at the OCUG next week.

A Contrarian Viewpoint on Container Networking

With DockerCon in Austin happening this week, I’m reminded of last year’s DockerCon Seattle, and watching some announcements with utter fascination or utter disappointment. Let’s see if we can’t turn the disappointments into positives.

The first disappointment has a recent happy ending. It was a broadly shared observation in Seattle, the media, and discussion forums: Docker was overstepping when they bundled in Swarm with the core of Docker Engine in 1.12. This led to the trial balloon that forking Docker was a potential solution towards a lighter-weight Docker that could serve in Mesos and Kubernetes too. Last September I covered this in my blog sparing little disdain over the idea of forking Docker Engine simply because it had become too monolithic. There are other good options, and I’m happy to say Docker heeded the community’s outcry and cleanly broke out a component of Docker Engine called containerd which is the crux of the container runtime. This gets back to the elegant Unix-tool inspired modularization and composition, and I’m glad to see containerd and rkt have recently been accepted into the CNCF. Crisis #1 averted.

My next disappointment was not so widely shared, and in fact it is still a problem at large today: the viewpoint on container networking. Let’s have a look.

Is container networking special?

When it comes to containers, there’s a massive outpour of innovation from both mature vendors and startups alike. When it comes to SDN there’s no exception.

Many of these solutions you can discount because, as I said in my last blog, as shiny and interesting as they may be on the surface or in the community, their simplicity is quickly discovered as a double-edged sword. In other words, they may be easy to get going and wrap your head around, but they have serious performance issues, security issues, scale issues, and “experiential cliffs” to borrow a turn of phrase from the Kubernetes founders when they commented on the sometimes over-simplicity of many PaaS systems (iow. they hit a use case where the system just can’t do that experience/feature that is needed).

Back to DockerCon Seattle…

Let’s put aside the SDN startups that to various extents suffer from the over-simplicity or lack of soak and development time, leading to the issues above. The thing that really grinds my gears about last year’s DockerCon can be boiled down to Docker, a powerful voice in the community, really advocating that container networking was making serious strides, when at the same time they were using the most primitive of statements (and solution) possible, introducing “Multi-host networking”

You may recall my social post/poke at the photo of this slide with my sarcastic caption.

Of course, Docker was talking about their overlay-based approach to networking that was launched as the (then) new default mode to enable networking in Swarm clusters. The problem is that most of the community are not SDN experts, and so they really don’t know any better than to believe this is an aww!-worthy contribution. A few of us that have long-worked in networking were less impressed.

Because of the attention that container projects get, Docker being the biggest, these kind of SDN solutions are still seen today by the wider community of users as good networking solutions to go with because they easily work in the very basic CaaS use cases that most users start playing with. Just because they work for your cluster today, however, doesn’t make them a solid choice. In the future your netops team will ask about X, Y and Z (and yet more stuff down the road they won’t have the foresight to see today). Also in the future you’ll expand and mature your use cases and start to care about non-functional traits of the network which often happens too late in production or when problems arise. I totally get it. Networking isn’t the first thing you want to think about in the cool new world of container stacks. It’s down in the weeds. It’s more exciting to contemplate the orchestration layer, and things we understand like our applications.

On top of the fact that many of these new SDN players offer primitive solutions with hidden pitfalls down the road that you won’t see until it’s too late, another less pardonable nuisance is the fact that most of them are perpetrating the myth that container networking is somehow special. I’ve heard this a lot in various verbiage over the ~7 years that SDN has arisen for cloud use cases. Just this week, I read a meetup description that started, “Containers require a new approach to networking.” Because of all the commotion in the container community with plenty of new SDN projects and companies having popped up, you may be duped into believing that, but it’s completely false. These players have a vested interest, though, in making you see it that way.

The truth about networking containers

The truth is that while workload connectivity to the network may change with containers (see CNM or CNI) or with the next new thing, the network itself doesn’t need to change to address the new endpoint type. Where networks did need some work, however, is on the side of plugging into the orchestration systems. This meant that networks needed better programmability and then integration to connect-up workloads in lock-step with how the orchestration system created, deleted and moved workloads. This meant plugging into systems like vSphere, OpenStack, Kubernetes, etc. In dealing with that challenge, there were again two mindsets to making the network more programmable, automated, and agile: one camp created totally net-new solutions with entirely new protocols (OpenFlow, STT, VxLAN, VPP, etc.), and the other camp used existing protocols to build new more dynamic solutions that met the new needs.

Today the first camp solutions are falling by the wayside, and the camp that built based on existing open standards and with interoperability in mind is clearly winning. OpenContrail is the most successful of these solutions.

The truth about networks is that they are pervasive and they connect everything. Interoperability is key. 1) Interoperability across networks: If you build a network that is an island of connectivity, it can’t be successful. If you build a network that requires special/new gateways, then it doesn’t connect quickly and easily to other networks using existing standards, and it won’t be successful. 2) Interoperability across endpoints connections: If you build a network that is brilliant at connecting only containers, even if it’s interoperable with other networks, then you’ve still created an island. It’s an island of operational context because the ops team needs a different solution for connecting bare-metal nodes and virtual machines. 3) Interoperability across infrastructure: If you have an SDN solution that requires a lot from the underlay/underlying infrastructure, it’s a failure. We’ve seen this with SDNs like NSX that required multicast in the underlay. We’ve seen this with ACI that requires Cisco switches to work. We’ve even seen great SDN solutions in the public cloud, but they’re specific to AWS or GCP. If your SDN solution isn’t portable anywhere, certainly to most places, then it’s still doomed.

If you want one unified network, you need one SDN solution

This aspect of interoperability and portability actually applies to many IT tools if you’re really going to realize a hybrid cloud and streamline ops, but perhaps nowhere is it more important than in the network because of its inherently pervasive nature.

If you’re at DockerCon this week, you’ll be happy to know that the best solution for container networking, OpenContrail, is also the best SDN for Kubernetes, Mesos, OpenStack, NFV, bare-metal node network automation, and VMware. While this is one SDN to rule and connect them all, and very feature rich in its 4th year of development, it’s also never been more approachable, both commercially turn-key and in open source. You can deploy it on top of any public cloud or atop of private clouds with OpenStack or VMware, or equally easily on bare-metal CaaS, especially with Kubernetes, thanks to Helm.

Please drop by and ask for an OpenContrail demo and sticker! for your laptop or phone at the Juniper Networks booth, and booths of partners of Juniper’s that have Juniper Contrail Networking integrations: Red Hat, Mirantis, Canonical, and we’re now happy to welcome Platform9 to the party too. We at Juniper will be showcasing a joint demo with Platform9 that you can read more about on the Platform9 blog.

PS. If you’re running your CaaS atop of OpenStack, then even more reason that you’ll want to stop by and get a sneak peak of what you’ll also hear more about at the upcoming Red Hat and OpenStack Summits in Boston.