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.
Comments
Very easily it's firewalled by the virtual router which whilst it's still a VM it's not the VM that's running the application/site.
Also by using the NAT/Port forwarding it's possible to save public IP's
Why would you need multiple private VLANs though? The price is 4.3 euro per VLAN not per instance you run. Unless you have VMs that need to be seperated from each other.
@Maounique
Would you offer an exchange of iwStack credits to credit for other services ? For example if I need to stop using iwstack but order a XenPower VPS.
@afornic Well that's what I thought as well, but @Maounique is saying they charge per VM on the virtual network.
We needed a setup like,
3 Instances for Mongo DBs in Replica Set (2 GB per Instance), they need each other IP address to talk, having internal, IP address and port which cannot be accessed by outsider
3 API servers Instances (internally load balanced, 1 IP), talk to Mongo DB Replica Primary Master in local network
3 Web Servers for private user (internally load balanced, 1 IP), talk to Mongo DB Replica Primary Master in local network
3 Web Servers for public (internally load balanced, 1 IP), talk to Mongo DB Replica Primary Master in local network
It is about 6 instances, we need to pay (6 + 3 + 3 + 3) * 4.32 Euros as an additional fee, where as out of 12 ips, only 3 public IPs we need. Other IPs we want to be internal, should not be accessed from outside.
Azure don't charge for internal network except if we use their load balancer.
To me, we don't use any specific iwstack load balancer, VPN for private network, but technically we pay for internal ip :-) . I think, iwstack to revise this virtual network stuffs.
Yes I don't get the logic between charging per VM, Depending on traffic volume even a Small instance (1vcore 512Mb ram) could run a router
That's weird, so connecting a 1GB VPS to the private network has the same cost as the VPS itself?
To be honest it was so cheap I didn't even notice! Thats bad policy though, we pay for internal IP more than external IP! I guess one private network = one system VM as well. I think they've made a mistake here.
They should revise this policy. Ours is not really a hobby/personal project, we use it for business, we have use cases that needs internal ips. Virtual router/VPN is something else altogether, can be charged on usage basis, Virtual Lan may not be the case.
It shouldn't matter what it's for, but I'd urge them to look at what it actually costs them to provide and see if there is any room to change the pricing structure.
Yes they need to charge for the "router" as obviously it's using resources, but I could have 1 VM pushing a lot of traffic and 3 more doing next to nothing so charging per VM doesn't really work out.
IMO treat it like another instance, maybe if they need to charge for traffic at some internal rate, although if it stays within the same HOST anyway then it shouldn't matter to much.
Complex scenarios like vyala described above is what people use Cloudstack for. That's their target audience, not people just running a few VMs. I am running 4 VMs, one for nginx, two for MariaDB Galera replication and one for mail, all communicating through internal network, and only the necessary ports are open from the outside, for example the MariaDB ones are not accessible at all.
I believe this pricing policy is an error at their part, unless for each VM you connect to the private network another system VM gets created for network, even though I don't believe that's the case.
Linode also doesn't charge for local network, they lack the tools that iwStack has though, like firewall, port forwarding, etc.
If the internal network would be separated from the virtual router, we wouldnt charge either, however, EACH internal network needs own router, otherwise they would be shared, many people offer internal network, but that is not isolated, meaning instances of other people can share the same vlan.
I agree it is not an ideal charging scheme and we have it on the todo list for some time, but we did not consider it a priority, because people which need it have the means to pay, the ones who cannot afford it, can setup some firewall rules and allow only a certain IP to connect to a certain port and keep all VMs on the internet. Really isolated networks with VPN access only are needed in complex setups and the virtual router is needed also for other things such as load balancing, but that is beyond the needs of the regular hobbist which cannot afford 4 Eur a month per VM.
Perhaps when we will have a shared internal network in the low-end "cloud" with local storage, we will review the charging scheme, but it is not easy as it is a custom code in cloudstack and whmcs, it will be revised after we squashed the most annoying bugs in cloudstack itself, such as race conditions which generate stuck instances at a rate of 2-3 a day at least and failed snapshots which remain hanged and have to be removed from the database. Such a bug does need a fairly elaborate set of conditions to happen, but it is very annoying for all parties involved. We hope most of those conditions are removed now that we moving to more a suitable storage scheme and less and less instances remain on the old one.
Yes each virtual network needs it's own router, but that router could be serving several VM's so why charge per VM connected to the ONE virtual router.
Edit unless it's a virtual router PER VM (or at least per node) and they form some kind of L2 Tunnel?
We agreed already it is not an ideal case, so, Uncle made some time and today or over the week-end will reduce the VR price 3 folds, from 0.006 to 0.002, but will still be charged per VM.
Also, the larger instances will be discounted.
@Maounique, thanks to you & uncle, 1/3 of the price is better.
@Maounique - Thank you it's refreshing to see a provider that actually listens to customer feedback.
Out of interest is there any way to do private network's without the cloudstack "virtual router", I don't even care if you still have to charge for that but it would be nice to instead be able to connect the VM's to an instance thats on both the private and public network but using the routing software of a users choice. I.e in my case I'd probably use Vyos as it's what i'm familiar with.
You can add 2 NIcs, one on the public network and one on the private one. Or, you can use some kind of VPN over the internet.
The first thing should be better, you can create your own router that way you, by any chance, the cs vr is not good enough for you..
Just got mail about price reduction from iwstack.
BASIC 2GB
from 0.012 to 0.010 iwCredits per hour (-16.66%)
(from 8.64 to 7.2 per average month)
BASIC 4GB
from 0.024 to 0.020 iwCredits per hour (-16.66%)
(from 17.28 to 14.4 per average month)
BASIC 8GB
from 0.048 to 0.040 iwCredits per hour (-16.66%)
(from 34.56 to 28.8 per average month)
BASIC 16GB
from 0.096 to 0.070 iwCredits per hour (-27.08%)
(from 69.12 to 50.4 per average month)
VIRTUAL ROUTER
from 0.006 to 0.002 per hour per instance (-66.66%)
Thank you for listening.
That's great news - it's really excellent value now. Looking forward to rolling out more 16Gb instances!
No love for us who have smaller instances?
Nope
The smaller instances are already well priced compared to DO et al. The 16Gb needed adjusting as it only increases to 12 cores (not 16) from the 8Gb/8 core instance.
We use one of that big instances for some website at work.
I wonder if they will have something with even more cores, or something like adding cores for an extra price, because this website is a little bit nasty on cpu usage =/ and the Dallas instances have slower processors than the xen instances.
Hi @Maounique what about Dallas iwstack instances? You many time suggested to use iwstack (dc2 and Milano zone) as primary but with replica in "other datacenters". Is Dallas good enough for "other datacenter" ?
i really did not use it at all and i have just started using it for slave of my primary website and everything is good.
can you please share uptime of dallas and what major issues were there?
thank you M
S.
You mean the 16gb ram will not be available any more?
Thx
"Discounted" from "descuento" (spanish), not as "discontinued".
claro! gracias
I am not sure if that is the description I gave.
Outside Milano, we do not control the datacenter, so, when shit happens, we are at the mercy of the people there. There were issues with US regarding the routing, more than once, when they changed something, forgot to announce our IP space and did not bother to check when we raised the issue, blaming our test sites for using poor quality carriers instead. I would say the total downtime was more than a day in 2014, possible up to two, but it did not apply to all locations over the world, only about some 70%.
At least, in Milano, Uncle can go and solve the problem there himself.
regarding the larger instances, we consider adding bigger ones. Bigger instances should have no problem in the Xen zone and the new KVM one where we are adding larger servers with better CPU. The old KVM zone has 128 GB ram per server, putting up 32 GB instances will probably not fare well there for various reasons, besides, the number of cores will be limited because, if one instance runs amok and locks cores, everyone will be impacted if all the cores are shared by it, it might even make the node unresponsive creating a cascade of moves and shutdowns, but Xen will manage this kind of things way better.
Besides, more than 16 GB needed, that should probably be on a dedi unless the advanced features of iwstack are needed.
@Maounique what happens if you run out of iwstack credits?
does your data get deleted or do you get given time to rectify the situation?
Although Hopefully you get a warning that the balance is running low anyway.
Yes, you get a warning 7 days before based on current credit and usage.
Your account gets suspended but you continue to get charged for storage and IPs, where storage involves snapshots, volumes, ISOs, templates, whatever takes space on disks. After a while, the data gets deleted, but no earlier than a week in which time negative credits add up, so, if you wish to recover your data but do not wish to use the service anymore, you need to add credit as soon as possible. This is not doable without opening a ticket, add-ons cannot be bought for suspended services.
Sounds more than reasonable.
I moved my FreeBSD instance to IWStack Xen and I have to say I should have done it earlier. The instance is much faster than the old one, the apps are loading quicker and overall responsiveness is much better.