New on LowEndTalk? Please Register and read our Community Rules.
All new Registrations are manually reviewed and approved, so a short delay after registration may occur before your account becomes active.
All new Registrations are manually reviewed and approved, so a short delay after registration may occur before your account becomes active.
Looking for beta testers (giving free resources)
Hello there!
Since a few days we started to tinker around our infrastructure and released our first version of vCDN, currently we are looking for beta testers in order to test:
- CDN Unmetered Caching/Ratelimiting/Firewalling
- Edge storage
All our features will be for free until we gather some feedback from the users
If you are interested or you just want to try it out feel free to reach me out
I hope i am not on the wrong section
Comments
Do you support IPv6 origins similar to how Cloudflare proxies things?
if it's a reverse proxy and they have ipv6 then it should...
Correct! I only ask because bunny does not so apparently its not as ubiquitous as I would like, and I don't feel like signing up for their service if it doesn't have IPv6 because it wouldnt fulfil my needs for such a product.
That said, my next obvious question is where are your PoPs/servers? You should have a nice global mix if youre hoping to be a top tier CDN provider.
oh, they do now
@vcdn Would be interested in trying. How to contact you ? DM ?
They do? Can you be more specific, did you check?
@vcdn I would be interested in trying, if ipv6 on edges.
joined Jan 2, no provider tag? Is it allowed to data gather here? Ive seen some did this before so I'm not sure if it is allowed. But IMHO it shouldn't.
No AAAA records (ie. no v6) were found.
@vcdn: Could you tell us where your edge storages are? Hopefully your CDN has some location that spans the globe?
If you got them I'm interested in testing.
@vcdn: You're not selling but you make an offer, so I think you need a provider tag. Read here:
https://www.lowendtalk.com/discussion/153228/rules-for-selling-on-lowendtalk-updated-october-2020/p1
Hi Erisa
At the moment we didn't enable dualstack compatibility on our ingress, We can run some tests with you if you want to try it
At the moment we are running beta tests with a single PoP in Germany, later on we can expand based on user requests
We cannot afford to pay upfront colocations or many nodes for a testing phase (as any cheap provider in the world)
Feel free to DM me or reach out our site here https://vcdn.it
You should be able to sign up
We are still in a testing phase and i wasn't able to post here so here i am
Feel free to contact us in order to make some tests
how long is the test period ?
Sorry, but I dont know how you want to create a CDN if you're not testing multi-location support during beta. There are some points you have to think about at this point.
For example:
It sounds like you don't even thought about this points. And thats just what I know about this.
For now your "CDN" is just a static filehosting service located in germany, sorry to say.
At least 1-2 months
no i just made that up
https://bunny.net/blog/ipv6-returns-to-bunnycdn/
You are perfectly right but as per our infrastructure and use-case, testing on one PoP is fine because we have hyperconverged nodes.
Edge storage will start only on one region then later on we will enable replication on the cluster and begin tests with compatibile S3 clients
I'm experienced in designing network protocols so I'd ask some deep questions.
When the client writes a file and the system returns "success", what's the semantics of this success message?
Does the system allow overwriting an existing file with a new version?
If so, how does the client know that the new version is available at all locations?
Suppose an earthquake occurs and the region hosting the primary copy of the storage bucket is destroyed:
Any slot?
@vcdn: think about the points that been mentioned! I'm not sure, but if you don't have the money for 3 nodes in your beta (which only lasts 1-2 months as you said), why do you think people will pay you for your service in 1 or 2 months, with one PoP? Or do you release with more locations? Remember customers are not waiting for a service they already can buy at bigger and sure better CDN-Providers with many locations over the globe, so there will be a long time until you maybe will get some ROI and that only if you have multiple locations...
Hello @yoursunny
Thank you for your questions, i am going to elaborate the answers here:
1) When the client writes a file and the system returns "success", what's the semantics of this success message?
When a client writes a file into our systems via APIs we return a specific status when the transaction is complete, if you use our dashboard and the upload
fails you might get a traditional HTTP error code
2) Does it mean the file is written to the primary only? (in this case, if the primary fails at this moment, the file could get lost)
Our infrastructure currently writes only on the primary nodes, if this fails at the moment of the transaction you require to re-upload it.
Before the end of the beta we will do some failover stress tests regarding data integrity
3) Does it mean the file is written to the primary and a small number of replicas?
Does it mean the file is available in all locations?
We are still designing the replication system so it's still an ongoing design, there are many options here but all depend on latency tests.
At the moment the files are not replicated
4) Does the system allow overwriting an existing file with a new version?
Yes we do allow overwriting existing files
If so, how does the client know that the new version is available at all locations?
As explained before, still an ongoing discussion but that shouldn't worry us since will be taken care of the fs clustering system (e.g Ceph, Openstack Swift... or the best solution that might fit into our nodes)
5) Suppose an earthquake occurs and the region hosting the primary copy of the storage bucket is destroyed:
a. How does the system find remaining copies of a storage bucket?
We will try restore them from the surviving servers it's the same if an entire AWS region burns all the local S3 are lost unless you have extra-replicas offline or on another S3 region/provider
We could plan offsite backups for all premium users (or when the replication system is ready use it at our advantage to restore data)
b. How does the system select a new primary region?
Still not decided because of replication design ongoing
c. If the network is partitioned, how does the system ensure only one region is selected as the new primary?
Same as above
d. If each side of the partition could select a different primary, how does the system reconcile conflicting updates when the partitioning heals?
Same as above
Great clustering questions, but if we do not use an active master-master replication those last scenarios are not an issue, if we do use an active master-master we will use the replication software raccomandations to ensure HA.
As an example if you write to a Cross-Region replicated S3 bucket there is no "master-master" write replication, instead it is an unidirectional async copy between 2 or more buckets
Hi @sbittmann
Thank you again for your feedback but for this beta we will give out only one PoP in order to narrow down all the issues and ease the troubleshooting
We will release more locations as soon as we have more beta users, currently in our point of view it doesn't make sense to scale up to N regions if you have a very few users testing
@Lutung
Sent you a PM