Dropping Cloudflare for Bunny.net
jola.dev
jola.dev
https://www.ftc.gov/business-guidance/resources/disclosures-...
An acronym as common as GDPR.
Or there is a loosely defined locally-run thing called 'Trading Standards' which is done at the council ("municipality") level.
and for the record I am just being difficult and everyone in tech/mildly well read knows what the (U.S.) FTC is. My point is more that one country's rules don't always matter for the operations of domestic commerce in another amongst their own citizens.
We famously mock our own jusrisprudence - "if Parliament passes a law that it is illegal to smoke on the streets of Paris, then it is illegal to smoke on the streets in Paris", so even when hard legislation exists (4chan/Ofcom shitshow?) it is meaningless.
The only power that matters long term in the universe is sheer force and hard power, and it has always been that way.
Not yet. Working on it, though.
If they can't they probably should move to an international focused site.
Either way, we are on the internet. Pretty international stuff.
In a time where more people usually beg for forgiveness instead of asking for permission, it already has
But that changes things
I don't like free offerings, because what if they decide to charge someday? What if someone decides "free is not feasible, we start charging $20 per instance now".
I'd rather have a low fee now, a change from $2 to $3 is more likely and that's fine for me. But from free to not free is risky for me.
I also like smaller, independent-ish ompanies that actually care about developers. That's why I use bunny.net, transistor.fm, Plausible Analytics.
You can just move to another provider at that point. At least when it comes to CDN and DNS there’s literally no vendor lock-in.
You can grab your dns records export them to csv and import somewhere else easily and a CDN is just a file server so you can just give your files to someone else easily.
1. For DNS we have standardized AXFR requests which the DNS provider needs to support as they are part of the DNS standard. There is not an option of not having that unless you have a really shitty provider that you should change anyway.
2. Same for Mass Import because again DNS already defines these things at the protocol level.
And resetting 2FA or whatever is just the cost of using any service
Personally I have used CF for ~10 years so I have saved $240 and I simultaneously use GitHub Pages and CF Pages for CDN because again I just need to give them a bunch of static files. Adding a third CDN provider would literally be a single command at the end of my build pipeline.
> Minimum Account Balance
> In order to keep your service online, you are required to keep a positive account credit balance. If your account balance drops low, our system will automatically send multiple warning emails. If despite that, you still fail to recharge your account, the system will automatically suspend your account and all your pull zones. Any data in your storage zones will also be deleted after a few days without a backup. Therefore, always make sure to keep your account in good standing.
You proactively replenish your balance, so in the worst case, you can just let the account go.
And how is that related to me? My comment said (and the parent I replied to) mentioned DNS and CDN.
Now we add compute services, data storage, whatever D1 is and the other comment mentioned auth/authz
Are people not aware what CDN and DNS are?
ehhhh, really depends on which CDN features you're using, and at what volume. Using ESI? VCL? Signed URLs or auth? Any other custom functionality? Are you depending on your provider's bot management features which are "CONTACT FOR PRICE" with other providers? Does your CDN provider have a special egress deal with your cloud provider?
It's possible to picture this being easy in the same way that being multi-cloud or multi-region is easy.
I have no idea what two of those acronyms mean. None of this is part of what a CDN offers.
Yes if you use DDoS protection, or cloudfare’s ZeroTrust or embrace $X proprietary features then what I said no longer applies.
I strictly said DNS and CDN.
VCL = Varnish Configuration Language i.e. how you configure your Fastly services
If you're just using a CDN as a proxy then there's no lock in but plenty of sites are using CDNs for much more than that
Practically, any metered supplier can put you out of business. It usually doesn't happen because destruction is mutually assured.
+1 for using smaller, more independent companies in any case!
This logic doesn't hold much water, however. Abrupt changes in pricing or other conditions happen with paid tiers as well
But if a free offering suddenly says "We are getting rid of free, only starting $899 a month baseline, because we noticed our free users aren't converting and we only want to support enterprise from now on". Well, then I have to move everything.
Still a big price hike can come, but +20% monthly is easier to stomach than if I can't be sure what will happen to the free offering.
With free offerings, you’re always helping the supplier in some way. Then you become the product. Which makes it difficult to understand the value exchange; it’s much easier to do so when you’re just paying a fair sum of money.
Same thing can be applied to low fees services. Look at bitwarden as a recent example, they doubled their prices overnight and hid the announcement in some useless blog post that no one reads, users found out later they are expected to pay doubles in any plan.. why? Because “that low price isn’t feasible we start charging $23 plus tax for premium now”
And the same things you will do to migrate to another when free -> paid you will do when paid_low -> paid_high if the pricing isn’t up to your expectations.
There’s no protection against this, it’s up to the business model change, investors, greed (aka scale up!), you name it, so regardless whether you are paying now or not, always have other options ready for when you need to abandon the ship.
But I hit a real issue recently: CDN edge caching served stale HTML after a deploy, and the service worker cached the bad response. Took a CDN purge from the dashboard to fix. The debugging experience when things go wrong at the edge is painful, you're always guessing which cache layer is the problem.
That being said, the free tier is hard to beat for getting started. Workers, Pages, KV, R2 — you can run a full production app at near-zero cost until you hit scale. Not sure if Bunny offers that.
Some of you may be skeptical about this but it allows for much easier management when working on multiple SaaS/hobby projects/personal tools.
As a consequence I've had to build quite defensively - adopting a PWA approach - heavy caching and background sync. My hope is that latency improves over time because the platform is nice to work with.
There’s a cost limit to how much high availability is worth on any project but vendors like CloudFlare don’t respect that.
A week ago I (a hobbyist running a small side project for a dollar or two a month in normal usage, so my account is marked as "individual") got hit with a ~$17,000 bill from Google cloud because some combination of key got leaked or my homelab got compromised, and the attacker consumed tens of thousands in gemini usage in only a few hours.
Google denied a rate adjustment, and haven't reached back out to me for a good few days now. My credit card denied the charge because it was over my credit limit by a good few thousand dollars and they suspected fraud, but now I am terrified of being taken to collections and ruining my prospects of renting an apartment due to my credit score/history being ruined, or them just taking me to court.
I am never going to use "use now pay later" services, especially with cloud portals where it's so hard to put in a actual cap, and the cloud provider not having any sane rate limits. I am fine paying if it was negligence or a mistake on my part as a very expensive lesson in security, but 17k is brutal.
The fact they don't have an easy way to hard cap usage (especially for an individual account) and have ineffective rate limits (how on earth is an account that pays a few dollars a month able to run up tens of thousands in just a few hours), makes me never want to use their (or any use now pay later with no easy caps or rate limits) service ever again. Or even a phone number to call.
Also before doing this save anything important that Google owns (gmail, youtube videos, anything in storage). The leaders at Google are vengeful enough to completely lock you out for challenging them.
That being said, I had enough issues with Bunny and CF debugging across regions that I made this free tool to do both remote HTTP and TCP traceroutes to keep my sanity: https://dnsisbeautiful.com/global-http-availability
That said, the edge-caching being how it is, it's possible to run into some race-conditions where the cache has been purged but not propagated to the edge network, and if visited too soon, the stale version might end up back into the cache.
[0]:
curl --fail --output "/dev/null" --silent --show-error -X POST "https://api.cloudflare.com/client/v4/zones/{our_zone_id}/purge_cache" \
-H "Authorization: Bearer $CLOUDFLARE_CACHE_PURGE_API_TOKEN" -H "Content-Type: application/json" \
--data '{"hosts": ["instorier.com", "www.instorier.com"]}'If i see something horrific like:
import * as BunnySDK from "@bunny.net/edgescript-sdk" BunnySDK.net.http.serve(async (request: Request) =>
Thats a proprietary lock-in worse than what it tries to replace!
> You provide handlers that fulfil requests from the system.
As I said previously, though I wish they were, such handlers are not part of WinterTC.
And then again, how those handlers are registered is also not part of WinterTC, which I also wish it were.
> APIs like that leak implementation details
How?
Almost all runtimes, like Bunny Edge Scripting, Cloudflare Workers, Deno, Bun, etc. use the same basic signature for the handler:
(request: Request) => Promise<Response>
Only how you register said handler is, unfortunately, different for each runtime.
The point of this discussion is that you can self-host, and you have a good chance of migrating the code away entirely. That's a big benefit that isn't "an attempt to get free labor". For that use, not only does it not matter if it's meaningfully open source, it doesn't matter if it's open source at all.
But in practice, we almost never receive major contributions from outside the team. Which is fine. We're happy just to have our team working in the open.
The reasons we open sourced it are:
1. Support a realistic local dev environment (without binary blobs).
2. Provide an off-ramp for customers concerned about lock-in. Yes, really. We have big customers that demand this, and we have had big customers that actually did move off Cloudflare by switching to workerd on their own servers. It makes business sense for us to support this because otherwise we couldn't win those big customers in the first place.
[1] https://developers.cloudflare.com/workers/runtime-apis/handl...
I assume by Service Worker API, you mean FetchEvent? [1] As in:
self.addEventListener("fetch", (event) => {});
That is a web standard indeed, but not one that was designed for use as a general-purpose HTTP handler, but rather for use in a Service Worker proxy.
[1] https://developer.mozilla.org/en-US/docs/Web/API/FetchEvent
I really loved the idea of using service worker sdk and having something that works in browsers too so the conceptual model of edge functions are spanning backends, edge POPs and browser runtimes.
I hear this argument all the time, but I think it's more complicated.
Firstly, if people used more diverse / smaller services the distribution of outages would change. While there will likely to be more frequent "smaller" asynchronous outages, many platforms can still break even when only one of their dependencies break. So, you might likely to face even more frequent outages, although not synchronous.
Secondly, we are not sure if these smaller services are on par with the reliability of Cloudflare and other big players.
Thirdly, not all Cloudflare infrastructure is fully centralized. There is definitely some degree of distribution and independence in/between different Cloudflare services. Some Cloudflare outages can still be non global (limited by region or customers that use certain feature set, etc).
If you actually care for the resiliency necessary to survive a provider outage you should have more than one provider.
Which means you should be running your own origin and using the simplest CDN features you possibly can to make your use case work.
> If you actually care for the resiliency necessary to survive a provider outage you should have more than one provider.
Well, that probably means duplication. This might be too expensive in certain situations. Also some occasional outages might be not a big concern for some, such as most bloggers.
I'm not downplaying the downsides of centralization. Certain things should be decentralized if it's reasonable. But it's not always that way.
A few people here are complaining about the lack of a free tier, but Magic Containers can cover a lot of the same ground as Cloudflare's Durable Objects, which IIRC cost a minimum of $5/month.
I feel like you missed what the author meant with that phrase. The author wasn't talking about for their website, but the internet as a whole.
> I can’t help but feel that the idea of centralizing the internet into a single US corporation feels off.
The point of picking Bunny.net is that it's alternative to this single entity that's got so much of the internet running through it, and is less susceptible to the BS in the US.
The solution was to move to Bunny, and that worked for everyone.
Not many DNS management providers (that I'm aware of, please correct me) support CNAME flattening. That is having your A record point to a CNAME.
Every time I purge the pull zone cache, I do it twice, cause once from my CI isn't enough. My CI does individual page cache invalidation during deployment, but there needs to be some kind of delay (with no feedback) when assets are distributed across.
also I said this in a another thread, they charges 1$ even for single testing http request.
Almost all technological choices I made as a teen were driven by "what hosting can I get for free, as my parents sure as hell won't put down their payment information for that". Back then that usually meant PHP and a max. 50MB MySQL.
In my case, and it was the 90s, I took the time to setup a way to pay by calling a premium (1-900) for $1.49 number so the barrier to entry even for kids was still reasonable.
Maybe in modern day the equivalent is adding Google pay and Apple pay then you cover some kids at least (gift cards and such).
Quite the hassle for the provider, and it will turn away any person who cares about privacy. There's no way to win anymore.
Good riddance to the "free" model. It's never actually free. You either pay with your data, or have to consume ads, or you're forcing other customers to pay for your free usage.
?
So 1 euro a month is too expensive for you? Wow.
Just pay the 1 Euro or go to GitHub where that is free but goes down almost every week.
Highly recommend their Edge Containers product, super simple and has nice primitives to deploy globally for a low latency workloads.
We connect all containers to one redis pubsub server to push important events like user billing overages, top-ups etc. Super simple, very fast, one config to manage all locations.
Bunny bills per resource utilization (not provisioned) and since we run backend on Go it consumes like 0.01 CPU and 15mb RAM per idle container and costs pennies.
But, a few things could be more straightforward. Cloudflare makes the whole static site and DNS zone piece feel far more seamless. With Bunny you will still need to stitch records between different parts of their dashboard.
edit: I'm thinking of the use case where the CDN as a proxy for APIs and uncachable content as well, where it used as a reverse proxy for transit/ddos protection.
Granted cloudflare also does DDOS protection, and that makes sense for an API. For that you could do some DDOS protection without stripping TLS, but it can only protect against volumetric attacks like syn/ack floods and not against attacks that are establishing full TCP connections and overwhelming the app server. (rate limiting incoming connections can go a long way, but depending on details, it might still be enough to overwhelm the serving resources, your use case is up to you to understand).
At some level, it's like they become your edge router.
https://social.mikutter.hachune.net/@mok/116208294430782702
BunnyCDN intentionally mis-writes any Mastodon request signing, as to make it incompatible with Mastodon.
And, they confirmed it's intentional.
The user sent in a help ticket, and Bunny confirmed this response rewrite was intentional and would not fix it.
I wanted to get this out, not to conjecture as to why.
Does anyone have thoughts or disagree on this in terms of pricing and cost effectiveness?
Im still learning the whole thing myself and plan to play with it this weekend and use it for a client site. Feel free to ping me again somehow in the future to see how it turned out.
I’ve now been with Bunny.net for over a year and have been very happy with the service.
https://www.goodboydigital.com/pixijs/bunnymark/
I'd assume most bots don't have a GPU attached :)
Easy upload of bind test files Flattened CNAME to support naked domains Robust free role based permissions to add other ppl
Anyone have suggestions for moving a stack of domains, many being little community and hobby projects away from cloudflare for a small overall price. Agency pricing like migadu offers for email on custom domains is what I have in mind.
50 cents per domain per month 10 cents per million queries
That’s prob cheap enough to support lots of little hobby sites and bigger traffic sites likely have some budget.
[1] Not completely sure but I think this was the incident https://blog.dnsimple.com/2020/07/incident-dns-resolution/
https://www.cloudns.net/premium/
https://www.luadns.com/pricing.html
I've found every other offering to be lacking. Some examples: Cloudflare is alright but has settings footguns if you're not used to Their Way of Doing It™ (e.g., before using DNSControl, I had to manually flip switches to turn off proxying every time I updated my zones). deSEC is free and okay, but sometimes quite slow to propagate and its UI+API are unwieldy. DNS Made Easy is often pushed on social media, but it's ridiculously pricey for what you get if you don't need a SLA. DNSimple seemed nice but IIRC I couldn't get a different API token per zone (?).
I'm currently relying mainly on LuaDNS. For me, it functions as a "dumb" DNS host (i.e., not using their Lua configuration-as-code system). Their API is oddly designed, but it's been passable since a recent-ish update, which has allowed me to safely port my zone files to DNSControl.
Annoying for dynamic DNS and DNS ACME challenges where you want a server to manage its own records and nothing else
We overhauled our account tokens a few years back: https://blog.dnsimple.com/2023/11/scoped-access-tokens/ . With account tokens you can specify fine-grained scoped access control, including specifying only one or more zones that a token has read or read/write access to.
Some of our users were unable to reach our CDN altogether. They couldn't load any assets at all. Bunny's customer service was far too slow to respond and mostly gave unhelpful answers. They couldn't even identify the issue.
In less than 45 minutes, I moved our CDN entirely from Bunny to Cloudflare Workers. Now our CDN just actually works, I don't have to debug our CDN for the Bunny customer service team.
Also, this is obviously a marketing post.
For DNS, it's cool because you can have it return scripted results which makes dns-based load balancing a bit "better".
The CDN certainly has it: https://bunny.net/blog/ipv6-returns-to-bunnycdn/
Depending on where I query from, OP's blog does have it as well:
# host jola.dev
jola.dev has address 37.19.207.38
jola.dev has IPv6 address 2400:52e0:1a04::1310:1I have IPv6-only backends and I had to select serving from the main POPs rather than the entire network (which is fine by me as they are also cheaper).
Only using edge storage, DNS, and CDN so far but very happy with Bunny.
I still have no idea what any of this has to do with any clients moving from Cloudflare to Bunny.net, what am I missing?
It's ridiculous.
They recently upgraded the player for streaming media, we use in one instance for tutorial videos, that apparently adds some missing accessibility features. All we needed to do was adjust the embed URL structure we were using and all set.
Pricing link is the second one in the top menu, and it shows all prices in a straightforward way.
See https://bunny.net/gdpr/. Also noticed this:
While uncommon, bunny.net also provides a way to block users from the EU from accessing your content altogether by using our traffic manager tools if you do not wish to serve users from the European Union. Which I assume can be reversed, only serving to users from the EU.
>One of my biggest concerns though is around how easily I could become heavily dependent on this one single company that then can decide to cut me off [...]
How does switching to Bunny make a difference?
It would be super nice to have a setup that uses multiple CDNs w/ automatic failover.
Doable, but that removes all the free tiers of all the CDN's. AFAIK they all require an enterprise account to keep using ones own DNS and their own GSLB DNS failover. There are probably a few exceptions and one could maybe make something of that but I don't know which ones are the exceptions.
The line between those two things in the case of the EU is awful blurry.
The Espace Léopold issues laws that are binding on member nations, wields significant power over trade, fiscal policy, and mandates open borders between member nations. These are hardly the features of a purely economic treaty organisation.
lol that ship sailed a long time ago it's certainly not a full federal republic but it's a lot closer to one then a mere "economic alliance".
It's not perfect but it's better than the alternatives and we really need a power bloc (even if currently only economic) that isn't the US and China.
Alternatives to US big tech are always welcome.
In my experience the TLDR generally gives the motivation/experience in brief. This feels more like a subtitle.