I can't wait for PaaS prices to drop. I know they'll always be relatively expensive, but I'm primarily a dev (not ops) and I don't have the time nor desire to manage my own servers securely.
Heroku lately has seemed a tempting offer, assuming I can run my apps on it (Rust based), but at $7/m for little side projects it felt.. expensive for a no user side project. They can add up. A $5/m DO box can host quite a few apps in containers, by comparison.
Perhaps Serverless would be a great playground for low/no traffic apps, but I've not gotten into serverless much.
That.. is really impressive! Had no idea about this, thanks!
For a well rounded review, can you or anyone comment on similar offerings with competitors? I was not familiar with per-request containers, but it sounds like exactly what I wanted.
I use Cloud Firestore, it's also serverless and is really easy to use for building SPAs as you can do a lot of things with just frontend code and trigger cloud functions when things in the database change.
I do wish there was a serverless SQL option like AWS has, but nothing yet. Can always use Cloud SQL, but as others have mentioned you need to access it over a public IP (for now)
Not OP but I can setup DB of choice on an instance or group of instances and access directly from the Cloud Run containers. Our particular main stack is on MongoDB 3.4 so we deployed Atlas on the VPC and the containers access it just fine.
As far as I am aware, there is nothing like CloudRun where your whole App are changed at a per second / request intervals. ( Although No Private Network means Your SQL requires Public IP which is a big No No )
I haven't tried using it though, and my recent submission [1] where BaseCamp had problem with GCP and moved back to AWS.
Although a public IP address is required, you don't need to poke any holes in the firewall. As a result, it is not a problem from a security perspective.
Cloud run looks cool, but doesn't seem to support databases ? Unless it supports linked containers?
I like that it's a mix of AWS lambda scaling with more open environments... Though it's hard to read through the marketing... "Nearly instance availing from 1 to 1000"
Cloud Run user here. It does not support running a database as a Cloud Run service; you'd need to connect to a Cloud SQL instance. This is one area I really hope PaaS providers improve on in the future. The closest thing at the moment is Aurora Serverless on AWS, especially with their neat HTTP connection system. That being said, cold starts are even worse since they're now the start up time for your application in addition to the start up time for Aurora Serverless.
But, correct me if this has changed please, last i checked Cloud Run did not support private networks. So you SQL instance has to be exposed to the internet, correct?
Your Cloud SQL instance would have a public IP, but would not be exposed to the internet. All traffic is blocked by default. Only GCP service accounts with the right permissions can make connections to your SQL instance, or IPs added to a whitelist (not recommended).
Unfortunately FaunaDB isn't a SQL database, so it's in the same class as things like Cloud Firestore, AWS DynamoDB, and Mongo Atlas. Good datastores for serverless applications, but not so great for traditional applications migrated to serverless.
Does CapRover's Web UI offer any kind of 2FA? It feels a bit scary to have the keys to the kingdom just hanging out in on a public IP address, for anyone to brute force the admin password.
No, but that's mainly because I still have to manage the server running Dokku I believe?
edit: For context what I was already using was manual docker on a DO box. While I recognized that Dokku would make it more seamless and easier, my issue was managing and securing the DO VM itself. I didn't see how Dokku would help that scenario
I would suggest reading about Ansible and working on an initial collection of Ansible playbooks that might help on that. Check out https://www.thecloud.coach/ansible-crash-course. Of course nothing beats a dedicated devops but some playbooks regarding main security compliances might do a great job. Anyone correct me if I am wrong but I am seeing it as a good base practice for these cases.
Possibly. I use Serverless as well but I prefer to keep my options open when playing with a new project, than trying to force everything into the same tool. If he is lean to learn something, Ansible might be good for setting some servers up, but there are some nice other suggestions in the thread if he doesn't want to.
One-click for initial setup, but what about maintenance? I like DO, but those easy setups seem like a recipe for vulnerable machines. If it's complicated to maintain, it shouldn't be too easy to setup.
It's definitely not as simple as you have to maintain the server yourself. But being able to run multiple applications/containers on a server for the price of one dyno really makes up for it for personal projects!
If I were working on a project with a corporate budget then using Heroku or a different style of architecture like serverless would be a no-brainer
I do CapRover + FaaS on the small $5 droplet on Linode. I run a lot of random stuff out of it, with the only change from me being increasing swap from 512MB to 2GB (it's all SSD backed anyways).
"Linode Kubernetes Pricing
LKE pricing includes only the resources you consume – Linodes, NodeBalancers, and Volumes. Your cluster’s master services are free of charge."
So whenever they are out of their beta, you should be able to run a $5/month single node K8S "cluster".
I don't see PaaS as a realistic alternative except for the simplest of apps. One always eventually needs to tweak the web server to get things like websockets, brotli compression, fast serving of versioned immutable files, or something else. Or run several servers with vpncloud.rs between them. It's not price that is a problem, it's the lack of flexibility.
That may be true for some, but it's not true for me - currently.
Ie, if I could get a DB/FS + PaaS for $5/m I'd use it in a heartbeat. If I run into a wall where I need to manually tweak some fine detail, I can always fall back to normal deployments, right?
PaaS in concept fits exactly what many of us seek. Deploying apps without managing ops. Just because it's not infinitely powerful doesn't mean it's a reason no one would use a PaaS. PaaS fits a very real use case.
In my somewhat edge case though, hobby/side projects, I can't pay much for them. So I manage the ops myself currently because I can save money, but if Heroku was just a bit more cheap I'd use them.
I've been a Netcup customer for a year and a half or so, and I love them. Locations in Germany only, but for less than $10/mo (5.29EUR), you're not going to do much better than this on price to spec ratio. Performance is fantastic. I have the 8G/300G SAS rootserver for 8.99EUR (rootservers just give you dedicated cores). Note however that (similar to BuyVM), they are also extremely quick to cancel you outright if you don't pay in time.
There is (seems to be) no way to quickly spin up an instance for testing for new customers. After completing the order form I received an email stating they will check my address details and only after that complete the order and let me pay.
Yes, you go thru the auth process, pay, and then they email you VM details after provisioning. Fairly standard. I don't think they claim to be very "cloudy," so yeah you're not going to get on-demand VM's you can just spin up like DO/Vultr. Perhaps they're hourly billing is confusing or misleading to lead you to think that. And it honestly really isn't hourly billing I don't think, unless they don't bill if you turn the VM off (never tested that).
They're a provider I haven't heard of before. Following your link, what does "Billing period 6 Months" mean? Elsewhere on the page it says: "Hourly based billing. Tariff can be terminated at any time."
I would recommend staying away from OVH for anything important (e.g., a mail server). While the service is fairly reliable, when things do go wrong (like an IP network that you share with others get blacklisted because one of the IPs is part of a botnet), their support is useless. They do respond and are very polite, but nothing actually gets fixed and you eventually realize they are just wearing you down hoping that you give up.
OVH is pretty much useless for anything 'important'.
Worked with a company that had a large spend account with them as part of our VM/VPS offering which was built on top of leased bare metal where OVH was handling all our Roubaix VM/VPS. Without warning they capped every server to 10Mbps. Their explanation was not given to us individually but posted in french on their forums. We only knew what was going on because a francophone customer translated the message for us. We'd gotten lumped in with seedbox hosts and large CDN accounts.
The gist was they didn't want to cater to high bandwidth consumers, something they didn't evaluate as (# servers / BW usage) but by (account / BW usage).
They wouldn't even lift the bandwidth limit to transfer data off the servers. Not even for private traffic. 10Mbps ports for all of them. We had to offer our customers free service in another region or their data.... Choose one.... Just to try and limit the amount of data we moved off.
If they'd given us any warning we'd have been perfectly accepting of them not wanting our business. Whatever they're reasoning, it's a good reason to never consider them for anything mission critical every again.
A big caveat not mentioned for Digital Ocean is that they blackhole IPv6 email traffic. It took a while for me to figure out why it wasn't working since it wasn't documented in their documentation and there was no ICMPv6 response indicating this. Moved to Linode after that.
I wonder if the same caveat applies to other providers that have improperly implemented IPv6 by not providing a /64.
I don't agree with their policy, but it is difficult these days to run your own outbound mail service and not end up in the spam box.
I gave up and pay for Gsuite as an SMTP smarthost so my outbound email isn't tagged as spam. That allows me to still process the inbound on my server if I want.
Obviously doesn't help the privacy concern if that is your motivation for running your own SMTP.
If, however, cost is a consideration, Yandex is an option. They will host SMTP for your domain at no charge. https://yandex.com/support/connect/add-domain.html (assuming Russian hosting is okay for you)
Been running my mail server on Linode for almost 2 years now. Fortunately I've yet to run in any issues. Having setup all the stuff properly from day 1 probably helps (TLS, SPF, DKIM, DMARC, MTA-STS).
Do you send a fair amount of email? Where I had trouble was running an ecom site. We set up all the things you mention correctly. We never sent any unsolicited email, but I assume people might occasionally flag a receipt or password reset as spam, and then it was game over for our email getting delivered. No issues since switching to gSuite.
My only gripe with DO is the stability for some of their regions. Some of them have been solid and only have minor outages, but others like one of the ones in AMS has had intermittent issues. For me their most attractive product is the DO spaces and is a quick drop in for S3.
But in all I like their servers and I go between the guides from DO and Linode to setup services that I’m unfamiliar with.
I've been running some live services on Hetzner Cloud (https://www.hetzner.com/cloud) and it's been very reliable. One server blip in ~1 year.
2Gb (20Gb storage, 20Tb transfer) from €2.49 - there are a few different DCs now so you can hedge your bets. The higher spec ones are great for Dockerizing a couple of postgresql backup servers.
Their VPS ('Cloud') offer is relatively new, with the dedicated and server auctions being their historical staple.
P.S. Very Euro centric, so probably not so good for Americas/AsiaPac
For those who want to lease ~anonymously with payment by Bitcoin:
BitHost OK cpu and ram OK uplink overpriced
CockBox OK cpu and ram slow uplink OK price
HostSailor OK cpu and ram fast uplink OK price
VPS.BG OK cpu and ram OK uplink OK price
I've used BuyVM for years and they are quite reliable. Though, I had one VPS terminated (including all data wiped) because I didn't pay the monthly invoice within 1 week of the due date. They have one of the most strict timelines on invoices I've ever seen for a web service:
4,2,4 - Unpaid services will be suspended four days past the due date on the service.
4,2,5 - Unpaid services will be terminated seven days past the due date on the service.
That said, the $15/year package is a super excellent deal for a basic VPS and I've never had any technical issues.
I've been using BuyVM for about six months, they're the only VPS provider I could find with anycast IP support which is a neat feature to have. I think they're a pretty small operation and in my experience, pretty much all of these low cost VPS providers will terminate if you go a few days overdue on the invoice.
I've been with them for my private mailserver VPS for ~7 years now and the support in IRC and via the ticketing system has always been fast and friendly. The author talks about the KVM/slice setup in the post but they're still dilligently running all the openvz from legacy contracts. You can count on them being around years from now.
I followed the general outline of the ispmail tutorials at https://workaround.org/ispmail . Debian based, postfix, dovecot, mysql, postgray (spam mitigation), catchall for the entire domain and only access it from imap clients or ssh'ing in. Doing webmail seemed like too much work to keep things secure. I self signed all my certs since it's just for me and very easy to set up and maintain (unlike letsencrypt over the same time span).
I have had the same experience. Used them on and off for years and really like the service, especially for the cost. Just be sure to set a reminder to pay the invoice because they don’t do recurring billing (or at least didn’t last time I used them), and is you don’t pay your stuff is terminated.
Have used DO, Linode and Vultr (and some others) for 5+ years but Vultr is the only one that keeps having network problems every now and then.
I run "monit" on all instances that checks for connectivity every minute but maybe every few months or so, I get alerts of network reachability for a slight moment from Vultr instance located in Silicon valley region. It's cool that they accept Bitcoin for payment though.
The page's test of course doesn't test reliability over the years and it seems it's not done 10 times on different instances, and the performance could've randomly fluctuated on the test, so I'd take the results with a grain of salt.
DO has been pretty stable for me. It just keeps running with no interferance.
Linode sometimes (like once a year) makes your instance unavailable for a tiny moment for maintenance but they provide you with clear explanation beforehand and I think it's good that they actively patch security problems that way.
Also good that they pool bandwidth from multiple instances, so even if your small instance takes lots of bandwidth, if you got some other instances, it can consume the limit from those too as a single pool.
Another for Linode is that they release OS images so fast as in they had CentOS 8 image the day I read the news it was released. I checked the others but only Linode had it then.
I have been happy with binarylane hosted in Sydney.
I tried vultr ~2 years ago and at least then they did not have their own ubuntu repository mirror. All the package manager downloads were painfully slow. Digital Ocean and binarylane mirror the repo and any package can be pulled pretty much instantly and it makes the initial setup much more pleasant.
+1 for Vultr. Not in NZ but we've been using them for various app / database servers for coming on four years now and we have literally only ever had one minor issue. Their support is through ticketing only (AFAIK), but their support team is very fast at responding.
Just switched jacobsparts.com over from ec2 to vultr. Cheaper by the hour than a 3yr reserved instance on aws, none of the nickel and dime billing and the cpu is sooooo much better, cutting ~250ms off the response time. Had an unplanned reboot today. Got an email from support and then it went down for about 20 seconds. Overall there is tremendous value in the smaller hosts, and I loved the article; though doubt some of the hosts reviewed will be around for long.
Another important "potential pro" metric is how fast can you reset a server to a blank state without having to flat out destroy + create a new one.
This is super handy for testing configuration management tools.
With DigitalOcean they have a rebuild option where you can take an existing server and rebuild it with a new image. It takes like 10 seconds to wipe your server clean in 1 button press and you always get the same IP address. You also don't get charged for a new server since you're still using the original one.
It makes testing something like Ansible really pleasant because you can iterate so quickly. I know you can always test things in a local VM (and I do for most of it) but there are subtle differences in DO's base image vs a stock or bento vagrant box of ubuntu / debian.
I've been using Vultr for almost 3 years now. They had some problems with 1-3 server reboots a year, and I once had some trouble with IPv6 setup. The support was quite helpful, but they ended up moving my instance to a different host node.
They take Bitcoin (verification needed) too.
I received a good amount of credits for finding a security vuln in their CP (not severer), and prices are quite good to begin with. If anyone plans to use them, I recommend the High Freq line. Snapshots are currently free. Additional IPv4 and /64 IPv6 cost $3/mo. Network-level firewall and fast DNS (although NS in same /30, anycasted) is for free.
Block storage is reasonable and is on par with Digital Ocean, and is fast. I have 50GB $0/mo instance as an early customer.
If the person who did the testing reads this, thank you for sharing that information. I think it's important to know which providers are good hosting options for smaller companies and personal projects without providing a bad, unreliable or even shady service.
> I'm surprised that smaller companies have not moved away from static pricing yet.
The physical server has a fixed set of resources. If you let customers pick what resource they want to buy, you'll run out of one resource before the rest, leaving the server underutilized and increasing your costs. You can't sell KVM instances with no ram, no disk, or no CPU.
If you have enough physical servers and small enough customers, you can mostly solve this by carefully figuring out which customers to put on which machines. However, you've created a complicated jigsaw puzzle. And what happens if you need to shift customers between servers to balance things out?
I think Google is able to offer custom machine types because they're big enough, and because they're able to perform live migrations between physical hosts (most hosting providers can't do this), so the customer likely won't even notice if they need to be migrated.
Instead, most of these providers offer a few different SKUs to satisfy different use-cases, or they just pick one target market (e.g. lots of disk space for backups or lots of CPU resource for compute) and focus on that. A few offer networked block storage, which gives some more flexibility, but at the (potential) cost of reliability and performance.
> None in Asia though? You need it to be in Taiwan to service the region well!
There is https://www.serverhunter.com/ but really reviews like this are better because metrics like CPU-core and instantaneous disk speeds won't necessarily tell you about the consistency in network, disk, and CPU performance over a long period, and the supported features. It is not like AWS and Google Cloud where you can expect stable performance.
I'm replying to myself because I just managed to get a IONOS instance running in Kansas City (same distance from east/west-coasts) for low-and-behold 1€/month with unlimited data (18GB SSD and 512MB RAM). How is AWS/GCE going to compete with that?
One gotcha is that if you come from the EU you need to register via ionos.eu and there is a registration fee of 10€ for each instance!
Also looking at VPS in china now, Shanghai is pretty well placed in the Asia region: Edit: VPS is crazy expensive in China (10x-30x more expensive than IONOS and that is without data which also is 20x more expensive!), I suggest using AWS as backup to GCE over there, probably want to stay outside of the firewall in Asia if you use port 80, atleast if your other Asian customers are more important than your China customers to begin with.
Something is off with the prices globally right now, the US is like on a permanent firesale when it comes to non essential goods, but to live there is impossible!?
Happy to see Lunanode rated Really Good. My exact experience during more than three years as a completely happy customer. They don't stock any Alpine Linux, but fetching the install image and setting up is a breeze, and not once, ever, have I had an issue with any of my instances. And highly appreciated: I can clone an instance, download the qcow2 image, fiddle with it locally, upload, and put it back in production. Flexible, easy, clean, no nonsense, reliable, cheap, and absolutely recommended.
Don't know about support, though, since so far I have had no cause to call them.
chuckled a little when reading the first negative of luna was the absence of having a debian 10 template. i raised a ticket with them about this in july, and have been periodically testing new openstack releases from https://cdimage.debian.org/cdimage/openstack/current-10/ to see if the issue 'fixed itself' for some reason with debian 10, cloud-init takes around 6 min to raise network interfaces, which is why they haven't released a template yet. not sure if the issue is on lunas side or not.
Regarding Scaleway lack of rDNS for IPv6: it's because they don't support it. Customers (like I was) asked for it many times since 2016 but it seems too hard to implement for them.
In my experience Scaleway isn't worth the risk. I had multiple weeks-long periods for different vpses (and both FreeBSD & Linux) over the course of two years where their network-mounted root disks would start throwing errors and then drop offline entirely. Worse, after one of these incidents their control panel usually wouldn't allow a reboot. I went round and round with support and the most they ever did was clear a flag for that and tell me they were experiencing network issues. I've also used Ramnode, Linode, Servercheap, Amazon, Heroku, various small local colo companies over, etc. and Scaleway was the only actively bad one.
Unexpected outages, disk corruption, and usually not even a credit when I'd been up and down for days. If you have problems like that with your infrastructure and they crop up on and off for years, you need to bring in new people and invest in infrastructure. They aren't going to.
I have similar experience with Scaleway for the disks. I have been running a few cheap dedicated ARM/Atom servers there for many years. It seems be a tied to a specific host or model, I have one of the first arm's which has been dead stable, but I have seen disk issues with both my new Atom's and newer ARM's.
I still use them for some non-critical stuff because their cheap dedicated ARM and Atom's give you isolation like a dedicated machine. In combination with a LUKS mount I like to think it's safer against any Meltdown/Rowhammer-like attacks than a shared instance.
But if you want a stable VPS you might have better luck elsewhere.
IPv6 is a second-class citizen on Scaleway. eg. IPv4 addresses are independent resources that can be assigned to any server (or load balancer), IPv6 are tied to a specific server, and can't be moved to an other one. And their load balancer does not support IPv6.
Two related questions not answered in this article:
- Where can I find good low-cost KVM with decent monthly bandwidth in Australia? Primarily want a remote fast Linux/VNC desktop (which will chew a lot of bandwidth), and want bandwidth left over to actually do things with.
- There's no mention of Oracle's always-free cloud offerings (100GB disk, 1GB RAM, 10TB bandwidth). Interested to hear what people think of this.
Yes. Instances keep their public IPs for the instance lifecycle. The IP is not reserved though, it is attached to that instance. So if you destroy that instance and spin up a new one, you won't have that IP anymore.
The worst case is probably that their "free" service has fine print that results in their lawyers handing you a large bill, probably by making it easy for you to accidentally use paid features. (Precedent: VirtualBox extensions)
In addition to Linode, Vultr, OVH, there are some local home grown "equivalents" Binarylane and Quantum Core, they both have a presence in Sydney, with the former having locations in other cities in AU, BNE, MEL, PER.
My preferred remote desktop tool is TeamViewer. Almost no VPS actually supports this, because TeamViewer needs a virtual graphics card (that's what I'm calling it, but I actually don't know what to call it).
Linode used to provide one at 1024x768, but I think they no longer do so. Vultr however does support one, at 1280x1024.
I'm guessing what you're getting at here though is that you want to manage the server through a desktop interface, complete with windows and menus and the like. Most people don't do that on Linux servers because most of the things you need to interact with to administrate one are only accessible through a command line interface.
That said, if you really want to manage a Linux server graphically, you might be able to make it work with something like Xvfb.
Yeah, I want a desktop. I manage my server via ssh of course, any extra graphical cruft is a liability there.
> Xvfb
I might, but I've tried a zillion things, and they all suck when you compare it to the ease of use that TeamViewer offers. What I also left out, is that I've already got the TeamViewer client running, because I remote into macOS and Windows.
You should search your package manager, not Duckduckgo. On Debian-based distros it's simply called xvfb, on Arch Linux it's xorg-server-xvfb, in the RHEL ecosystem it's xorg-x11-server-Xvfb but might require enabling extra reporitories.
I’m not trying to be snarky, but I’m curious, what is it that you like about TeamViewer? I don’t run X on any of my servers, so plain old ssh (and mosh) work well for me. Have you considered the alternatives, perhaps tunnelling VNC/RDP over ssh?
TeamViewer hasn’t had the best track record[1] when it comes to security (compared to something like OpenSSH).
> ssh works fine if you're using it as a server, sure
woadwarrior01 only mentioned SSH as a transport to wrap RDP/VNC, both of which are remote desktops, just like TeamViewer. So combining SSH + RDP/VNC, you both get the security of SSH, and the remote desktop part. Windows even comes with RDP (maybe macOS also probably comes with RDP/VNC as well) so even less software to install/maintain!
Yeah, I got that tunneling was assumed. In fact, VNC is madness without SSH tunneling. However when you look at macOS, the only VNC client that automatically sets up SSH tunneling is Screens.
But now comes the trick; one of your machines is behind a NAT. And it's macOS, or Windows, or what have you.
No VNC client is going to help you here. With TeamViewer, this just works. There is no open source offering that comes close, to my knowledge, because of the need for a proxy service.
This is one of those areas where if the company selling this sort of product isn't maintaining modules for their service in ansible/puppet/saltstack/etc...., they're really doing theirselves a disservice.
I've had to fix the saltstack vultr integration multiple times when they break it. it's really annoying.
I may have messed it up, but last time I tried Lightsail (being a loooong time user of Digital Ocean), I set up a CPU-intensive workload and it stalled within a short period of time. Then I found out the Lightsail instances are t-based ones, so I ran out of credits?
I didn't see time4vps on here. They are the only provider I've seen with large storage vps's for a reasonable price. I use one to do all my cloud backups and I can still run afew services on top.
Hmm... their largest storage VPS, 8TB, is $1437 USD/yr, which works out to around $15/mo/TB. This is 3x more expensive than Backblaze B2 @ $5/TB/mo. It seems like you'd do better using something like B2 for cloud backups and getting a $5/mo Linode VPS or $2.50/mo Vultr VPS for your other services. I use both Linode and Vultr and like them both.
LowEndBox indeed has great deals once in a while, but it's impossible to navigate. You can't search by specs. If you want a cheap VPS/dedicated server with certain specs, try
i would not recommend to look for hosts there. Most (say 90%) of the hosts seen there are hosted on Colocrossing that actually owns both LowendBox and Lowendtalk (forum, thats mostly ran by volunteers)
just today it turns out many of the hosts (20 and still increasing) posted there just suddenly decide to close shop, all at once, with the same email from all of them. This isnt even the first time this has happened, but it is on a larger scale. (you can read more about this on lowendtalk if interested)
Reverse DNS is what maps an IP address to a host name. A PTR-record type is used to store reverse DNS entries. The name of the PTR-record is the IP address with the segments in reverse order followed by ".in-addr.arpa".
For example the reverse DNS entry for IP 2.3.4.5 would be stored as a PTR-record for "5.4.3.2.in-addr.arpa". That then points to "somehost.yourdomain.tld".
It's important for things like email, where the receiver can do a reverse DNS lookup to make sure your host is supposed to be sending email from "example.domain". Lacking proper reverse DNS, your outgoing email could be marked as spam. Or rejected outright for one not existing. It's also used for some forms of SPF entries (though that's discouraged).
Last, it's somewhat useful for logs if you have hosts in your domain that connect to each other. Though, usually, you're encouraged NOT to configure apache, syslogd, etc, to do reverse DNS lookups...for performance reasons.
