Via this WHT offer. ONVPS is currently doing a 50% off discount on their premium VPS packages — no promo code needed. Their VPS plans start at $5.50/month. Here is the direct order link. The plan “Vps256” comes with
- 256MB memory
- 8GB storage
- 110GB/month data transfer
- Xen
The servers are in Seattle WA at Wowrack‘s new facility, which are actually owned by the same company “Technica Holdings, LLC” according to the about page (which also hosts an impressive line up of management team). Domain created in December 2006 but they have only gone online since 2010. Their underlying virtualization technology is Xen (explained on their blog post), and their are using OpenNebula to manage the infrastructure + their customized control panel (which has a bug tracker). Would like to see how everything tie together. Certainly not your average OpenVZ/SolusVM company, and quite interesting at this price point.
Related Posts:
- 5 Reasons Why You Want a Low End Box - May 26, 2021
- Dead Pool January 2012 - February 2, 2012
- exit(0); - January 19, 2012
I wonder if IPv6 is available.
does any body know the ip address arange of this host, then I can have a test to check the speed
Maybe you can use this:
216.176.188.1
http://noc.wowrack.com/20meg.ftp
those are wowrack’s though.
All for more XEN hosts in the industry!
I have a 256mb VPS with this provider. The sad thing is they only have 64bit templates, and dunno why but they have pygrub on the 256md version and it takes a lot of RAM (or they have not configured it properly), I dont use the VPS for anything since I’ve prepaid for 3months. Here are some stats,
[root@xxxxxxx ~]# uname -a Linux xxxxxxx 2.6.18-194.26.1.el5xen #1 SMP Tue Nov 9 13:35:30 EST 2010 x86_64 x86_64 x86_64 GNU/Linux [root@xxxxxxx ~]# free -m total used free shared buffers cached Mem: 256 214 41 0 28 66 -/+ buffers/cache: 120 135 Swap: 511 0 511 [root@xxxxxxx ~]# cat /proc/cpuinfo processor : 0 vendor_id : GenuineIntel cpu family : 6 model : 15 model name : Intel(R) Core(TM)2 Duo CPU E4500 @ 2.20GHz stepping : 13 cpu MHz : 2194.500 cache size : 2048 KB physical id : 0 siblings : 1 core id : 0 cpu cores : 1 fpu : yes fpu_exception : yes cpuid level : 10 wp : yes flags : fpu de tsc msr pae cx8 apic sep cmov pat clflush acpi mmx fxsr sse sse2 ss ht syscall nx lm constant_tsc pni ssse3 cx16 lahf_lm bogomips : 5488.59 clflush size : 64 cache_alignment : 64 address sizes : 36 bits physical, 48 bits virtual power management: processor : 1 vendor_id : GenuineIntel cpu family : 6 model : 15 model name : Intel(R) Core(TM)2 Duo CPU E4500 @ 2.20GHz stepping : 13 cpu MHz : 2194.500 cache size : 2048 KB physical id : 1 siblings : 1 core id : 0 cpu cores : 1 fpu : yes fpu_exception : yes cpuid level : 10 wp : yes flags : fpu de tsc msr pae cx8 apic sep cmov pat clflush acpi mmx fxsr sse sse2 ss ht syscall nx lm constant_tsc pni ssse3 cx16 lahf_lm bogomips : 5488.59 clflush size : 64 cache_alignment : 64 address sizes : 36 bits physical, 48 bits virtual power management:And the I/O is very slow, I never get above 10 MB/s.
Right now I am using it just to check different nginx configs, nothing else.
This is one I/O test I did just now.
That doesn’t look good ^_^
So poor performance :)
This is very interesting. I just ran this in my Vps256 host and got this:
But I guess I’d have a hard time trusting the guy trying to sell me server space…
Sam, if you could let me know your IP address, or email support at onvps.com with your IP address, I’d really like to figure out why you’re seeing such bad performance.
Thanks!
Tres
And this is from one of our Level 2 hardware nodes:
I’m very interested in finding out what’s causing the performance degradation for your server. Please get back with us as so we can get it fixed!
Thanks for posting the data, Sam. It certainly isn’t what we want to see, but now that we know about it, we can get it fixed.
Tres
Thanks for sharing!
I’d be a bit worried that they’re running Xen boxes on Core2Duo hardware. Especially since Xen requires proper hard disk arrays to perform well. Most desktop-grade hardware systems just can’t hold up under the stress of Xen, especially when it’s only a dual core system (possibly with HT, but doubtful)
Although the Level 1 systems use SATA/Core2-Duo, we try to ensure good performance by having a lower VPS density on each hardware node. We’re not trying to hide anything — I’ve even published our VPS density right here in this thread. I don’t think that’s very common. (Please correct me if I’m wrong; I’d love to see some of the other guy’s density numbers to compare against ours — especially the OpenVZ providers.)
As I’ve written above, Sam42’s performance is definitely not normal. If he had written to support, we would have taken steps to make sure that it was fixed. That’s why we’re calling this our beta period; we want to encourage feedback from customers. The lower prices are our way of thanking early-adopters who can help us ensure the best performance of the entire system. Obviously that’s not working in that case… But we’re definitely interested in making sure that the system performs well.
Tres
well,I can not trust a company that on its website says, Textually, (you can check in the front page of the web) “Intel Xeon Multi Core Processors”
“RAID 5/10 Disk Array”. And the tests says (textually)
MODEL NAME : Intel(R) Core(TM)2 Duo CPU E4500 @ 2.20GHz
stepping : 13
cpu MHz : 2194.500
cache size : 2048 KB
physical id : 0
siblings : 1
core id : 0
CPU CORES : 1
268435456 bytes (268 MB) copied, 34.58 seconds, 7.8 MB/s.
Well, the evidence speaks for itself.
I can’t see that text.
In other side, they are honest :D WYSIWYG
In the section “whats included”, where are items, item 14 reads (textually) “Intel Xeon Multi Core Processors”, and item 15 “RAID 5/10 Disk Array”, I can send you a site snapshot, if you cannot see this… :D
There is nothing about I/O tests. I assumed your test published textually, as you said. But ok.
well, uhmmm, How to say that my comment does not “moderate”. :D. you really think a company will post a result as bad on your website?. I thought it was clear that I referred to the result published by sam42. I just say, in onvps say one thing and reality another, published by sam42.
I can not trust a company that lies. I only say that.
It seems you have your mind made up. We’re not the right provider for you. That’s fine. I don’t expect you to change your mind; however, I do take umbrage to the idea that we are lying. I’ve done my best to respond to the problems you’ve pointed out with the website. In addition I’ve made more information available here in this thread than you will get from most VPS providers. Period.
If you have found the right provider for you, that’s great. We’re very happy that your business needs are being met. We wish you well.
And to be perfectly frank, I have a hard time believing that there aren’t vested interests at play here.
If Sam42 is publishing results from an OnVPS server, I don’t understand why we’ve never heard about the issue before. What makes it harder to believe is that I’ve asked repeatedly for communication to investigate and resolve the issue and have not received any. We consider the numbers that Sam42 is getting to be an issue that needs to be resolved as soon as possible. That kind of performance is what we’d expect to see from an OpenVZ provider, overselling all resources to the point where the server is not functional.
We aren’t interested in overloading servers and making quick dollars. We’re in the business of establishing a long term relationship with a client. We’re not trying to bootstrap things with one overloaded server chugging in someone else’s colo; we have our own space & we’ve got a rack full of servers ready. And if we’re seeing that kind of performance on our Level 1 systems, then we’re going to change the hardware that we deploy. It’s as simple as that. That’s why the site says beta.
I’ve published results that conflict with the “OnVPS” results that were published by Sam42. And although I wouldn’t expect that to change anyone’s mind, I would like to point out that Sam42 has never contacted us about the performance of his server. In fact, we’ve not received any support requests regarding performance AT ALL. From any client. Ever. And since I’ve been posting in this thread, Sam42 has been curiously quiet.
I’ve been doing this for a while. I know when performance gets like Sam42 is “sharing,” you hear about it from customers.
So Sam42, if you are interested in getting the performance issue resolved, please contact us via support (at) onvps.com or, contact me directly at tres.wong-godfrey (at) onvps.com.
If you are seeing that kind of performance, it’s a critical issue & we’re going to get it fixed.
8GB storage? that means they stuffing something like 60 VMs on one 500GB hdd
Read their entry blog on I/O and performance? :)
http://blog.onvps.com/2011/01/11/onvps-infrastructure-built-to-perform
This just proves just because its XEN it does not means you get a better VPS compared to OpenVZ.
Who wants a XEN on old outdated hardware ?
Hi everyone, I’m Tres from OnVPS.
I just wanted to take a minute to answer a few of your questions and concerns.
Before I get into the details I’d like to address is the issue of overselling. We never oversell our systems. I’ll cover this in more detail below, but I want to reiterate that we do our best to provide a good service for a good price.
First thing’s first: our hardware.
Hardware
We have two classes of hardware that we run. Vps256 and Vps512 are Level 1, while Vps1G, Vps2G and Vps4G are Level 2.
Level 1
These systems have four SATA drives in RAID 5.
They are multi-core systems, but are not Xeon
They have 7.5 GB of RAM that can be provisioned
So, at the very most, these systems can be provisioned with 30 VMs
That kind of density is much better than other places I’ve worked at in the past, but it gets even better with our Level 2 products.
Level 2
These systems have 4 SAS drives in RAID 5
They have multi-core Xeon processors
They have 15GB of RAM that can be provisioned
So, in our worst case scenario, our level 2 nodes can be provisioned with 15 customers. That kind of density is much better than other VPS providers that I’ve worked at.
Network Latency Test
An IP address to test for latency is 216.244.76.213. That’s my personal VPS. No web content yet, but it should work fine for ICMP.
64 Bit Templates
The 64 bit template issue is something that we’re fixing right now. I’m putting together a whole new set of 32 bit templates & we’ll be provisioning everything under 4GB using 32 bit templates as soon as the update is complete.
IPV6
We’re still working on IPv6. As of now, we provide up to 8 IPv4 addresses.
False Advertisement (Xeon multicore processors)
Now as Fred points out, the website content is troubling. I have to say that I’m really disappointed that that is still printed there. We’ve gone around twice now with our web-guy to get that fixed. You’re right, Fred, I’d be very wary of the same thing. The little things matter a lot when you can’t sit down and look at the guy your giving your money to.
And I know that talk is cheap. So if you’re interested in getting on a Xeon system, I’ll make a Vps1G coupon for Vps512 price for the first three months. Just let me know & I’ll make it happen. (That’s a 1GB Xen VPS for $9 a month for 3 months, then $18 monthly for life.)
Please let me know if you have any other questions or concerns.
Thanks a lot for getting your concerns out & thanks to lowendbox for letting us address them!
Tres
Just to be sure, are you certain that it’s four drives in Raid 5?
Hi Matthew,
Yeah, every one of our production hardware nodes is a 1u system with four disks in RAID 5. You can see a picture of the backside of our first rack at http://blog.onvps.com/2011/02/28/onvps-datacenter
Tres
Ter : What is the coupon code for “1GB Xen VPS for $9 a month for 3 months, then $18 monthly for life.”.
Also, where is your AUP and TOS. Are they same as WOWRACK.Com?
Hi innya,
The coupon code is LEBVps1GFor9
There’s only 10 of those available.
And yes, you can use the Wowrack documents.
Thanks!
Firefox browser hang and crashed while ordering (I just type in user information but no payment information), somehow it created an account for me but it did not create payment of invoice option on their sever management portal. So, there is noway I can pay unless cancel and reorder or add server option. I opened the ticket with sales and billing to get some informaiton. No one has been answered me ticket for over 24 hours.
I just left a comment on their blog and I have no idea where it went. Not sure if Akismet grabbed it or it’s held for moderation or what happened. If it was held for moderation, something should have displayed to show where it went. If it hit Akismet, well real professionals know Akismet has issues and there are much better solutions out there.
From looking at their website, it seems to me that this company is more interesting in tooting their own horn than providing potential clients with specifics as to what they represent. For example, there’s lots of talk about their 10 Gigabit Backbone Network but yet no mention of who provides it and/or what their peer arrangements are. (edit: I found a listing here but please note that they’re listed as being in the same building. Nothing about actual connections, peering or bandwidth. Their network page is lacking in detail as well. Only two uplinks? That’s scary. I have 2 dedicated direct and 2 general house pool uplinks and I’m tiny.) The four datacenters that they promote looks actually to be three that they rent space from. The fourth one could be owned by them though I see no real evidence of such. (edit: Rented. Makes me wonder if they even own their own equipment?) There’s lots of big companies and their names being thrown around but I have a feeling that if I called up any of them, all I would get would be a “Who?”
Sure, looks nice to the average customer but lots of holes when I look at it.
The normal WordPress config puts the first comment of a visitor in the moderation queue. after your first comment being approved, you can comment normally. I think the same happens here in LEB.
Thanks but the “You comment is awaiting moderation” notice is in core:
http://core.trac.wordpress.org/browser/trunk/wp-includes/comment-template.php#L1356
For it to not show, someone had to remove it or create a theme that bypasses that check. It’s rude to leave someone hanging like that.
We’re using Typo as the blog backend since it’s Rails (The control panel and most of the supporting infrastructure is Ruby/Rails based).
I haven’t seen any comments from you. I cleared out a few spam comments this morning while looking for your comment, but it was all obvious spam.
The spammers seem to be able to post, so I’m wondering what happened to your post.
Thanks for letting us know about the issue.
Tres
LOL, I guessed wordpress :P
But there is no info about the blog software in the bottom n_n
Now I want to deploy that blog xD
Their bandwidth is from spectrum networks and allhostshop which is made of tier 2 providers like he and bandcon.
http://spectrumnet.us/our_network.php
http://www.allhostshop.com/network
look at their logo, they are still on beta.
So?
they’re still on beta, and if something bad happen? they will just say we’re still in beta and still testing something(hardware, software, etc), but something wrong and all data wiped.
Hmmm, how to say this… I think they are just honest.
Suppose a company “not in beta”, they are testing HW and SW too, and if something goes wrong, your data will be wiped too. (Virpus is my nearest example).
In other areas… Gmail was in beta for years, and a lot of people used it, then, just they modified the logo for a logo without the “beta” word, and nothing else changed. something like that.
Then IMO a better criteria to consider a company “beta” maybe the time running (maybe less than 1 year). Then, at least a half of the companies posted here is beta, because they are in that time range.
Perhaps a LEB points system? DNS in 2 separate DCs = 2 points, etc… (Hmm, I thought I’ve seen this somewhere)
How many points for anycasted dns with more than four or five hosts? ;)
Well, about the DNS, why not to use some of the free services around the net?
I like the free dns of he.net. I don’t know if there is some clausule for usage on business websites. If I had a hosting company, I would use some he.net. (Sounds a little bit like a TV commercial xD)
“beta” doesn’t mean anything.. as was already said gmail is the perfect example.
and to be fair if their systems did fail even if they weren’t in beta… still not their problem you didn’t have a backup. no it’s not a pretty situation but backups are ALWAYS client responsibility no matter what a host says they will do
5.5$,the first month or for life ?
Hi Neal,
The price is for life.
Please let me know if you have any other questions.
Tres
WOW! Sorry, I did not mean to cause a ruckus here, was just posting my numbers. And that was meant to be a simple stats post like many other posted by readers post on almost company’s blog entries. As for why I was absent, I did not know you were here or were gonna show up hence did not bother checking. I have submitted a ticket. I really did not bother using the VPS after the initial few tests and when my first ticket on a unrelated matter was not answered. I had got the VPS for a 3 month deal and I did not mean to spread false information or lies about your company. I appreciate that you are open and trying to solve the issue. Please don’t take the comments expressed here to be an insult.
[root@xxxxxx ~]# dd if=/dev/zero of=test bs=16k count=16k conv=fdatasync 16384+0 records in 16384+0 records out 268435456 bytes (268 MB) copied, 33.1643 seconds, 8.1 MB/s [root@xxxxxx ~]# dd if=/dev/zero of=test bs=16k count=16k conv=fdatasync 16384+0 records in 16384+0 records out 268435456 bytes (268 MB) copied, 32.6315 seconds, 8.2 MB/s [root@xxxxxx ~]# dd if=/dev/zero of=test bs=16k count=16k conv=fdatasync 16384+0 records in 16384+0 records out 268435456 bytes (268 MB) copied, 34.8777 seconds, 7.7 MB/s [root@xxxxxx ~]# dd if=/dev/zero of=test bs=16k count=16k conv=fdatasync 16384+0 records in 16384+0 records out 268435456 bytes (268 MB) copied, 32.6548 seconds, 8.2 MB/s [root@xxxxxx ~]# date Thu Mar 17 23:30:32 PDT 2011 [root@xxxxxx ~]# dd if=/dev/zero of=test bs=16k count=16k conv=fdatasync 16384+0 records in 16384+0 records out 268435456 bytes (268 MB) copied, 33.4451 seconds, 8.0 MB/s [root@xxxxxx ~]# free -m total used free shared buffers cached Mem: 256 253 2 0 0 141 -/+ buffers/cache: 110 145 Swap: 511 0 511I ran the tests again a few minutes ago.
For anybody doubting me,
For anyone not trusting my post, I am willing to create a account for you.
Thanks.
Sam42 is a customer. He has sent in a support ticket and we have verified that there is in fact an issue with the hardware node he’s provisioned to. I’ve verified those numbers. That hardware node has problems.
We’re very grateful that he’s let us know about this issue & will be moving all customers off the poorly performing node to another one as soon as possible.
In addition, we’ll be providing service credits to all customers affected by this poor performance.
I’m sorry for impugning your trustworthiness, Sam42.
I suspect it is their RAID5 that causes the I/O to be so shot, it is well-known that RAID5 is not a write speed champion at all.
The RAID 5 issue is definitely not in play here. Running iostat on that hardware node shows a maximum throughput of this:
Our others look like this (not even max, just some throughput at any given moment):
That’s just looking through a few minutes of usage, but there’s definitely a bottleneck on that node.
My biggest worry is why you’re using Raid 5 in the first place. Raid 5 has lesser performance and is known to be troublesome in large I/O environments over time — Especially when a drive fails. It’s much, much harder to recover. Furthermore, the chances of being able to do a live rebuild are very very unlikely.
Honestly, I wouldn’t be surprised if one of your drives are failing and your raid card is trying to make up for the lost data by re-calculating based off of the other two drives.
Simply put: Raid 5 sucks and shouldn’t be used unless combined with another Raid level (such as Raid 50 or Raid 50+e, Raid 51, etc…)
That’s interesting. Some citations would be brilliant.
I can say that personally, I’ve worked with literally hundreds of VPS servers in this configuration (I was employee number 1 at Spry/VpsLink waaay back when). We’ve tested the same exact hardware in RAID 5 and RAID 10 and seen no performance hit from RAID 5. That’s both in testing and in practice.
In my experience, there’s no additional fundamental safeties that RAID 10 affords. Data is redundant on two disks — I can’t see which two the data is on, but the odds of losing data are the same for each. And in practice RAID 10 never gave a worthy performance boost (in fact, it was pretty negligible).
I’ve been pretty happy with this configuration. It’s worked well for hundreds of servers and thousands of customers. It may not be the best way, but every choice has its tradeoffs.
http://en.wikipedia.org/wiki/Standard_RAID_levels#RAID_5_performance
Hi Orien,
Thanks for the link!
The RAID 5 write issue really comes into play in big databases where changes affect a few bytes of data and changes get made extremely quickly — not really your normal use-case for a VPS. At the point where a client needs RAID10 over RAID5, they probably need a dedicated system for their database.
This is not to say that we don’t want to provide the best performance for our clients — that’s probably my primary goal; however, the benefits of using RAID 5 outweigh the costs for most use cases. Getting 8GB of storage vs 5GB of storage means a lot more to most customers than getting RAID optimization. In your typical VPS use-case, most file transactions will be reads. Databases will be relatively small and won’t have thousands of writes per second.
At the end of the day, I’m not married to RAID 5 or RAID 10. But I can’t make my choices based upon a single performance metric. I have to weigh all the options & choose what I believe most clients want. That’s a difficult task since everyone has differing needs; ultimately, I have to trade off some of one Good for another Good.
Thanks again,
Tres
Hi Tres,
Have you had the chance to compare the performance hit and time it takes to rebuild RAID5 & RAID10 array?
I do agree with you somewhat that depending on usage, RAID5 can match RAID10 performance. But my experiences with it is small. Due to all the random writes virtualization server makes.
I think the point Matt and Orien trying to make is the overheads involved between parity and non-parity RAID topologies. It (performance) gets worse and worse over time. I think that 90% of the well minded VPS hosts would settle on non-parity RAID most of the time.
I’m not here to debate with you (I’m sure Matt and Orien is not neither) on your OnVPS’s company infrastructure. You have listed vast amount of experiences in VPS industry. I think we just want to see if there are more advantages (other than space) that RAID5 offers over RAID10 you can provide…. since you listed 4 drives setup.
Thanks for stepping up and taking the time to answer all questions. Wish you guys the best.
Hi Teril,
I really do appreciate the feedback from you guys. Feedback from potential customers is invaluable & I’m never going to ignore that. If the RAID 5 issue is something that’s going to make us more attractive, then I’m going to weigh that more heavily in my decisions.
VPS is an evolving industry. Back when I started, it was a completely different landscape with a very different target market. While there are still plenty of customers who are only looking at RAM/CPU/Storage per dollar, there are others that are more discerning about what they are getting. It’s these kinds of customers that we’ve designed OnVPS for — the customers who know when they are getting good service for a good price.
So, I will be taking another look at the RAID issue because of your feedback (after all, if I’m not listening to potential customers, then I’m bound to be wrong). Thanks a lot to all of you.
Regards,
Tres
They probably just try the same “trick” as Iweb.com and others who pick up their old junk of hardware and puts virtualization on top and calls it cloud servers :)
These guys are probably just running on single 500GB disk without RAID :)
When is your promotion end?
Hi innya,
The half-off promotion will be ending fairly soon. It will likely be over within the next few weeks.
Thanks for your question!
My VPS was migrated and now I have the following results. Best amongst all the hosts I am using currently.
I feel sorry now for posting the bad results. I did not mean to slander the company. Hope they do well with their business and continue with their honest support.
Sam43: Could we have your actual download speed from vps?
Thanks Sam42
Whats up with the bs=16k ? :)
Can you provide output of same test with bs=64k please?
268 MB vs 1.1 GB
I was using that because previously it would take a long time with 64k. Anyways, here is one with 64k
I have a small requirement: I have an iPhone app that needs to communicate with a small database once or twice per user session. Storage and RAM don’t seem to be a problem… I am, however, concerned about how much CPU time I would get for my web service. Can you enlighten me?
Hi Rolf,
Thanks for your question. I can say that we do provide access to two VCPUs in every deployed VPS, and that we do prioritize CPU by plan, so larger plans get first priority. That said, of any resource that typically CPU isn’t the bottleneck on a VPS node. Normally disk I/O becomes a factor first. Even though the CPU is being used for disk I/O, normally it’s the disk heads that cause the biggest slowdown — so CPU is something we normally have plenty to spare. I think that’s true for most VPSs — whether they’re OpenVZ or Xen based.
I’ve actually written my own iPhone app that uses
NSURLConnection connectionWithRequestandNSXMLParseron the flip side to push and pull updates from a web server. That app is using Rails on the back-end to handle rendering and reading the XML and a MySQL database for storing everything.The nice thing with the iPhone API is that it is pretty good about HTTP transactions. It does well on slow connections & fails gracefully if the server isn’t available.
Even if you were doing something with Erlang or Node.js, I think you’d find that the VPS would be quite adequate for your needs in terms of CPU cycles available.
Please let me know if you have any other questions!
Tres
The only thing that prevents me from using this VPS is the bandwidth, their performance/price is great though. Alot of other VPS hosts have upwards of 500 for the same price.
Hi James,
We bill for any bandwidth overage. Additional bandwidth is $0.10 per GB of transfer.
That may not seem competitive when you’re essentially talking about $27.50 worth of additional bandwidth included for $9.00 monthly.
I’m not sure who is offering 500GB of bandwidth for $9.00 a month, seems like a great deal. That means that the provider gets less than $0.02 cents per GB, per month. Just what they make to offset their costs for bandwidth alone. That doesn’t include the cost for power or for hardware.
Almost seems too good to be true…
I don’t know who is offering this kind of pricing, but I can say that one of the old time provider tricks is to offer “unlimited bandwidth” and then meter the traffic so that there is no way to get anywhere near the quoted available bandwidth. Again, I don’t know who is offering this; I’m definitely not trying to accuse anyone in particular of doing this.
So if it works out that you really get 500GB a month for that price, and bandwidth is your most important consideration, I’d say go for it. Sounds like a great deal.
Thanks very much for your feedback.
Regards,
Tres
Just a heads up we were double billed by onvps for our vps this month. Also something is wrong with their ssl cert on their management site.
They’re one of reliable vps provider, but its too bad not offering cheap vps anymore.
“OnVPS is no longer accepting new customers or new subscriptions for existing customers”