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 Cheap cloud VPS that is truly hardware and data center redundant.
I'm looking for something that is truly data center redundant so that my websites, billing, and MySQL never goes down due to hardware or data center issues. Seems a lot of providers call their service a cloud service but it's not in my mind because it is not immune to a data center going down. That happens a lot more than most people think. Sometimes due to DDoS, sometimes due to edge router or switch problems. Sometimes due to power or floods (Sandy) etc.
I don't want to debate about trying to located in a good data center with a perfect record. Some of the most expensive data centers with the best histories have had problems. I'd rather just have a true cloud service and not have to worry about it.
Not interested in Amazon.
Comments
If you want it to be fully redundant as you say, I'm not sure "cheap" really fits in this title.
As cheap as possible with those features. I don't need any kind of management but I want something that is less expensive than Amazon. Even if I had the budget I would not be interested in Amazon.
CloudVPS.com gets my vote with VR.org (Host Virtual) a very close second.
Netherlands is not going to work for me so cloudvps.come is out. I should have specified that it needs to be in North America. VR.com doesn't say anything about datacenter redundancy. Will have to look into them more.
I prefer "Inexpensive" to "Cheap" on this.
Redundant is something not cheap but can inexpensive.
Please die 'true cloud' nomenclature. It's like 'unlimited bandwidth'. It just doesn't exist. Invariably, you'll need guest-level replication of file and SQL. Anything short of that is wishful thinking.
As mentioned, if you want true redundancy, you need to setup multiple servers with multiple providers in multiple locations.
If you went with one provider, even if they had multiple location redundancy, what happens if their management software gets compromised or has a bug which causes all locations to go down?
So there is no such thing as data center redundancy? Where all the redundancy is built into the system. If I wanted to do my own SQL replication I would have done that but it's a PiTA for what I am doing. I would basically have to set up my own cloud.
I would rather that it was just baked in like it would be on Amazon AWS. I don't know how else to say it but to me that would mean "TRUE CLOUD". Cloud means all over the internet. Not just one server or one data center but at least 2 of both. A lot of these so called cloud providers offer redundant VPS on at least 2 physical servers with auto failover. If they have a couple datacenters nearby then each server could be in a different building and that would be close enough for me. I just don't want them in the same building. Ideally not on the same power, same backbone, same edge router, same switch. Not in the same building at a minimum.
That was the whole point of calling it cloud wasn't it? Putting everything in one datacenter is just a cluster and not what I would consider a 'cloud' so that is not what I am looking for.
https://cloud.google.com/products/compute-engine
Yes I like the potential of that. However, so far they seem to only be targeting developers and high end stuff with charge by the hour so it ends up being quite expensive for production stuff.
AWS EC2/EBS instances are only single DCs (availability zones). Outside of RDS which does master-slave replication underneath the hood, nothing crosses that boundary. Compute Engine is setup in a similar fashion. Even people with multi-AZ apps on Amazon are building their own DRBD/GlusterFS/MySQL clusters on top of the basic provisioning if they need that level of HA across zones.
The main limit is physics. Distance is a big factor. But even with intra-city DCs with 2ms between them, DRBD-style replication on WANs is finicky and requires some careful management to ensure you don't end up with split-brain. No provider would want to be responsible for weird support issues that crop up from that.
True cloud is an annoying marketing term. Real multi DC redundancy exists. But it is a series of choices of compromises between money, performance, availability. None of which can be made without knowing the individual business case. That's why the 'true-cloud' myth of hands-off-everything-is-fully-redundant-and-split-second-HA is an annoying snake-oil marketing gimmick.
The Amazon zones are a bit fuzzy because they never say specifically if it's one data center or a cluster of data centers in the same geographic region. I am aware of a lot of those things you mentioned. I would be willing to accept the tradeoff. If you can asyncronously sync MySQL over WAN with replication and files/folders with rsync then I don't seen any reason you could not do it by baking it right in. I don't want to have to set all that up myself but I am now looking at global IP and IP failover that some providers are advertising. However, I am still stuck doing the replication myself which I don't want to do.
Good luck in your search. Just FYI,
http://media.amazonwebservices.com/pdf/AWS_Security_Whitepaper.pdf
That said, they're not without fault as @lennierb5 said, software bugs can cause the whole thing to collapse. Bad EBS replication code took down the US-East region not too long ago by a cascade failure.
Nothing is 100% but redundant data centers (with sufficient physical separation if in the same metropolitan area) gets you pretty close. Actually I am surprised how often we have trouble with data centers. We are using some pretty good providers and are willing to pay more and have found that is no guarantee. It's very hit and miss trial and error. Also like I said even the very best data centers and providers have outages sooner or later.
To me the best solution is to not pay extra for supposedly better data centers and just make that variable redundant.
Linode NJ and Atlanta is my preference. Can't get much more HA and redundant than that.
https://blog.linode.com/2013/03/07/linode-nextgen-the-network/
Possibly you can find something like this, someone may build it for you. Get ready to pay a lot
Most of the clouds are VPS anyway. The things that tend a little bit more to real clouds are atleast clustered VPS. I am yet to hear about Datacenter redundancy, especially for less than Amazon's offering.
Having A + B power feeds in a facility, that has diverse fiber networks, comming from multiple locations, having UPS and generators with enough fuel, fully redundant network..should be way better than having your stuff in 2 datacenters across the street, still vulnerable to big natural disasters, power cuts in the region and network cuts.
Thanks for the suggestion.
I don't see what they are doing that is all that different then any other decent provider or am I missing something? They have pretty charts that show redundant cabling to 2 routers/2 switches which is just standard data center practice. They also seem to think that because they use the word "Cisco" it is supposed to mean they are great when Juniper infrastructure is arguably considered better. I know of one data center that recently tore out their Cisco equipment and replaced it with Juniper.
I don't have any bias against them and took a good hard look at the informaton on their website. However, after scratching the surface I don't see anything that excites me all that much. Do they use OnApp for their cloud infrastructure?
This is what I see from that Amazon AWS document.
"Data centers are built in clusters in various global regions. All data centers are online and serving customers; no data
center is “cold.” In case of failure, automated processes move customer data traffic away from the affected area. Core
applications are deployed in an N+1 configuration, so that in the event of a data center failure, there is sufficient
capacity to enable traffic to be load‐balanced to the remaining sites"
That pretty much sounds like exactly what I want. I just don't want to use Amazon. Partly because I don't like their AWS control panel and also don't like their pricing and also don't like the fact that latency and ping times are reportedly highly variable. Probably due to overloading.
CloudVPS does offer data center redundancy across 3 data centers but there is no geographic diversity. All 3 DC's are in Amsterdam (Equinix, EU Networks, Global Switch).
Host Virtual also offers it with their private cloud dedicated HA node but setup fees are $2400 and monthly fees are $2400. Quarterly payments only so almost $10K to launch.
GoGrid (and I'm sure other enterprise level cloud providers) also offers it but they are much more expensive than CloudVPS or Host Virtual.
The startup and monthly costs of a HA cloud solution with data center redundancy far outweigh the amount the average LET provider is likely to lose if their site goes down for a few hours (or even for an entire day).
Yes, CloudVPS looks good.
This statement on their website says it all.
"Your data will be stored in at least three different racks containing
storage machines. We guarantee that at least two different data
centers will be used to store any piece of data."
That's what I want. As long as it will work with MySQL databases and I'm not exactly sure if it will. They are using OpenStack object store so I guess I should search for providers in N. America doing that.
@sman Sure redundant network, rock solid hardware, instant support, and just launch one vps in each location. Blows away any LEB provider. Configure mysql replication or what your web stack needs. Welcome to hardware and data center redundancy at an afffordable price.
With truly redundant data centers you should not need rock solid hardware. You don't even need RAID. The 2 data centers are essentially the RAID
@sman sure if you like to increase your risk and likelihood of possible future annoyances. Your looking for something exotic that doesn't exist at affordable prices.
CloudVPS appears to be offering a VPS that does it at 10euros a month + Object store pricing which looks similar to Amazon S3. I don't need much storage. So doesn't seem that exotic to me. Or is CloudVPS that much further ahead of everyone?
@sman try it out, id love to see you review it up. It looks like you're SPoF would be the SAN
We're using Cloudvps. Site was offline for a while as the network was under DDOS. I do not think this is what you are looking for.
CloudVPS is great. I've been a customer of them in the past and the company I work for has several servers there. The DCs are not next to each other, but also not too far away (I think 20km at the most).
They really know their stuff. They've got several in-house (real office, not a bedroom) Xen experts working on their custom fail-over software, IIRC. It's finished, of course, but there's always a new version to be worked on. I even believe one of their people contributes to Xen.
Why isn't anyone doing what CloudVPS is doing in N. America? Besides Amazon of course.
About the only full proof solution to DDoS that I know about is having sufficient bandwidth to absorb it. It takes even the best DDoS equipment 2 or 3 minutes to detect and null route it from what I have read so that is not the full solution
@sman Because there's already Amazon? Nobody wants to compete with them in that space.
Full proof solution to DDOS... i dont think one exists... and bandwidth to absorb 300 Gbps (eg. Spamhouse level) attacks is expensive.
I guess Amazon might be my only decent option still. I hate their interface. A lot of things that should be easy are made way too complicated.