Subscribe
Sign in
Porter — Heroku that runs in your own cloud
Porter is a platform that makes AWS/GCP as easy to use as Heroku. With instant deploys from Git, built-in autoscaling, and automatic SSL, Porter gives you the convenience of a PaaS while preserving flexibility and control.
104
Replies
Best
@nazim_m we actually already use Terraform under the hood and can automatically provision to AWS, GCP, and DO. You can also just connect an existing k8s cluster. Happy to follow up if you have questions about a specific set of providers!
I’ve been fortunate to be one of Porter’s early beta testers and I gotta admit — they probably have the best PaaS solution, both operation- and capacity-wise, as well as cost-wise, especially if you have AWS credits. 100% would recommend trying out!! 🤩
@bcye Our hosted dashboard is just a window into applications/services running in your own cloud account. While you can run a local version of the dashboard itself, it's a bit of a pain to get spun up so we're currently working on significantly improving the local installation process. The honest answer regarding pricing is that it's subject to change in the future - as of now, we don't even have billing set up. Bottom line though, our goal is not to charge indie devs and small teams, but charge larger teams with features that are geared more towards collaboration.
Been using Porter for a bit now, they've made the process of spinning up EKS clusters in AWS dead simple. Their UX is very well designed and easy to work with. Congrats guys!
Hi everyone! 👋 Thanks @mwseibel for hunting us.
Our goal with Porter is to provide the best parts of a PaaS like Heroku while keeping the benefits of a mature cloud provider like AWS or GCP. We believe most startups shouldn't have to worry about cloud management / DevOps. Heroku-like platforms are great for this: a repo goes in one end and a hosted application comes out the other. Unfortunately, this automation tends to come with a hefty price tag, and as you grow (and inevitably need more control), you're locked out of the underlying infra since you're hosting in someone else's cloud.
Porter deploys everything to your own AWS/GCP cloud, so you can always go under the hood when necessary (and even use your cloud credits 💰).
Here are some key features:
- Deploy directly from a Git repo or a Docker registry
- Built-in SSL
- Built-in autoscaling
- Zero downtime automated deploys
- Add-ons and applets for databases, logging, analytics, etc.
- Runs on Kubernetes and can connect to an existing cluster
- "DevOps mode" for advanced users (includes a graph view of resources)
- 100% open source and MIT licensed
Any feedback would be greatly appreciated! If you're interested in tracking our progress, we'd love if you could watch/star 🌟 our repo. You can also join our Discord community and follow us on Twitter.
P.S. We're hiring!
@mwseibel @justin_rhee1 not only does Porter make K8 as easy to work with as it should have been. But the team is amazing. You really get the feeling that they're creating the product with you and it translates to a level of confidence that I'd typically only have in more mature companies. Well done guys!
Come for the easy to use infrastructure on your GCP or AWS credits. Stay for the team and the amazingly designed product.
Hey @justin_rhee1! Nice tool, although I have to ask, will you only support "big" cloud providers? Or do you plan to extended to others (like my beloved Linode)
Thanks @andre92! Currently, you could provision a cluster using the Linode Kubernetes Engine (LKE) and then link that to Porter via the CLI. Hope that helps!
Looks great! Kinda specific question here: What types of built in metrics are there? Also, if I want to see if my users are seeing 50x before requests reach my app server (e.g. at the AWS load balancer or nginix ingress level), can I tell within Porter or do I need to roll my own observability for that? With other K8s solutions out there, having visibility into requests before they hit app servers has been lacking and leaves me worried we'd have errors we're not getting metrics for.
@scott_ames_messinger thanks! We use Prometheus as the built-in metrics solution, exposing some pretty basic defaults: the rate of network received bytes, the memory usage bytes, and the CPU usage (for each metric, you can view it per-pod or summed across all pods in a deployment). Since we install the NGINX ingress controller by default, you would be able to view these metrics for NGINX, but not at the AWS load balancer level.
We're experimenting with different metrics views since they're pretty easy to implement with Prometheus -- for example, for the NGINX ingress controller which exposes custom metrics, we could display things like upstream latency, number of connections, etc. Feel free to reach out with more specific questions!
@abelanger Are you all using the AWS NLB or ALB? If the ALB, we can see those request metrics in the AWS console which would be fine. If the NLB, having NGINX metrics would be vital. Otherwise, we're just flying blind and NGINX could be throwing 502s and we're none the wiser. If you all add those, please let me know! I'd love to migrate over.
Also, I didn't see any documentation about this, but I'm assuming it's easy to rollback to previous versions (like in heroku) and environment management is done through the web ui?
@scott_ames_messinger We support both the NLB and ALB based on the ingress annotation, but default to NLB pointing to the NGINX ingress controller. This setup seems to be the best user experience for custom domains due to the static IP address that gets assigned, so I think we're going to be primarily supporting NLB + NGINX in the short to medium term. Will definitely reach out when we have custom NGINX metrics!
In terms of rollback, yes, you can roll back to any previous version of the application. For environment management, we don't have the equivalent of Heroku pipelines or preview environments (yet!), but we do support multiple clusters per project that you could use for staging/production.
Porter is a must-have tool for someone who wants to build scalable apps without having to invest a lot of time figuring out and learning cloud infra and dev-ops
Shoutout to @sunguroku1 at Porter who has been more than accommodating in helping us deploy our app on K8's despite my lack of proficiency in it.
@khu, myself, & friends have been on porter for a few weeks now. Great experience for small teams and founders are superstar people! +1 a hundred times over
@khu @taehyoungjo Thanks Tae and Kevin! Looking forward to your launch soon :)
@sandochee great question. The main difference is that CapRover is built on top of Docker Swarm whereas Porter uses Kubernetes under the hood. Either works, but we're definitely seeing more widespread adoption of Kubernetes as the orchestration layer, especially in recent years.
I haven't even installed it yet, but I think this is something I've been wanting that I didn't know I needed. Def going to be trying out later!
They have the hands-down best PAAS product out there: the best crafted UI, the most smooth on-boarding experience, and the best customer service that will put a smile on your face. The founding team is such a incredible collection of talent!
I wish something like Porter existed when I started my first software engineer job at a 1 year old startup, where I spent too much time doing DevOps work + trying to navigate through the complexities and quirks of Kubernetes and Amazon EKS, rather than working on important features. Being able to simplify EKS cluster deployment and visualize + configure helm charts via GUI would still be extremely valuable today even if I am already familiar with Kubernetes and Helm. The fact that Porter is cloud-agnostic and lets me keep control of my deployments also solves a key reservation I had with other Heroku-like platforms. Can't begin to express how excited I am to give this a try!
@tpow Thank you! We had the exact same experience at other jobs/startups. Feel free to reach out if you encounter any issues!
Does porter support the creation of provider-maintained databases/caches like AWS RDS, Elasticache or GCP Cloud SQL?
Wow! Such a great product. I will definitely try this with my digitalOcean cloud.
Porter acts like a mothership to all the k8 clusters is very intuitive in managing the work load and the deployment 👍
Congratulations to @justin_rhee1 and the entire Porter team. They've done an amazing job helping us transition to Kubernetes via Porter. We were spending a non-trivial amount of time managing cloud containers - provisioning, scaling, monitoring metrics, etc. All that's gone ever since we switched to Porter!
This looks amazing @justin_rhee1 . I'll give it a spin over the weekend for sure! Great to see it's OSS as well.
We have been using Porter for a couple of weeks now. Launching and managing instances was effortless and extremely quick with the help of the Porter team and it allowed us to focus all of our efforts on developing our product rather than learning Devops. Awesome team and very grateful for their continued support and suggestions to maximize the efficiency of our hosting practices
koinju