Web Hosting Talk : Web Hosting Main Forums : VPS Hosting : Ubiquity hosting review

[BTamlan]

I signed up with Ubiquity hosting in June. I was looking for an unmanaged VPS solution, and I liked that they had a Chicago data center since Chicago is closest for me.

The server I'm hosting is fairly basic, just a personal server for use by myself and my family.

* apache, mysql
- used to host some personal web sites (a few WordPress blogs for family members, fan site for a band) that don't draw a lot of traffic -- averaging 700 hits/day, 226 Mb/day.
* sendmail & dovecot IMAP
- only 2 mail users including myself.
* nagios
- used to remotely monitor some servers at my day job.

Total bandwidth for August was 16 GB in / 6 GB out for August, so the bandwidth usage isn't intense.

Sales:

The guys at Ubiquity were great as far as the sales process goes. Very helpful, even switched the installed OS for me at no charge when I changed my mind on what OS I wanted.

Service/Support:

For the first few months, Ubiquity was great.

A few weeks ago (Sept. 9), I started having some problems. Most of the time, the server was just ridiculously sluggish. Apache, sendmail and IMAP connections would time out. Any commands typed at the command line took several seconds to respond. At first, Ubiquity said there was nothing wrong. Then they decided it was a problem with another user on the node running an IRC script. They said they warned the IRC script kid to knock it off, but performance never improved.

About a week later, the performance got even worse -- now, it was bad all the time, not just most of the time. Finally, on Sept. 17 Ubiquity says that some disks in the raid array my VPS is stored on failed. A week later, they've finished rebuilding 4 of the 11 disks in the array and the server is still unusable -- still getting timeouts on apache, sendmail and IMAP connections.

Support-wise, for me, having an unusable server for 2 weeks and counting is a deal-breaker. I'm already in the process of moving moving my server elsewhere (but I'm not done yet and have not requested cancellation yet, so Ubiquity guys, if you're reading, please don't pull a GeekStorage and just up and delete my stuff.)

The support from Ubiquity on the above described problems was inconsistent. When the problems first cropped up, as I said, they pretty much just said they didn't see any problems on the node, but could move me to another one. Then they changed the story to the IRC script explanation and just closed the ticket without any further explanation (for example, "so, at this time we won't move your VPS to another node."). I repeatedly asked whether the usage I described above was within the capabilities of their VPS solution, but never got an answer. At some times, I'd get quick responses in the middle of the night. Other times, questions went a day with no reply.

So, bottom-line, good sales process, and when things are working, they work well. But when things aren't working well, it seems Ubiquity isn't able to rectify them in a timely manner and/or reliably communicate with customers about the issue(s).

I'm now hosting with Future Hosting. I'll be back with a review of them in a few months.

[qwidjib0]

The sluggishness mentioned relates to this..

http://support.ubiquityhosting.com/i...news&newsid=33

I have not seen any VPS that have been made unusable for even a day because of this, but to be fair I don't see everything happening on every one of our servers at all times. I do apologize for the IO performance issues while this runs, however we do not need to rebuild the RAID array in a modular server every day (and the good news being no data was lost.. and no downtime occurred). If you have a more timely method of restoring the RAID array in a modular server, or have suggestions for how we could have made that announcement clearer, I would definitely be interested in your input so we can improve in the future.

[devonblzx]

It took 3 days to rebuild just two drives on the modular server?

[qwidjib0]

Yes, IO load escalated very high on this system, and is taking roughly 24 hours per drive to rebuild.

[Jay H]

BTamlan,

Have you had an account with us in the past? If so, please provide your ticket ID that you requested cancellation and we just "up and deleted your stuff".

If you've never had service with us before, I would suggest you not drag our name into your misfortunes.

Sorry to hear about your troubles.

[BTamlan]

Originally Posted by Jay H

BTamlan,

Have you had an account with us in the past? If so, please provide your ticket ID that you requested cancellation and we just "up and deleted your stuff".

If you've never had service with us before, I would suggest you not drag our name into your misfortunes.

I was half-jokingly referring to a recent thread in this forum ("Disappointed in GeekStorage,") where one of your former customers says he gave you a heads-up that he would soon be canceling his service and you went ahead and canceled the server and deleted his data before he was able to migrate it elsewhere.

Kidding aside, I'd sooner put up with more of Ubiquity's "no really, it's not technically unusable ... you can still login, right?" shenanigans before i'd sign up with a host that would do that to a customer. (Even if yeah, the smart thing to do would've been to wait until after you've migrated your data off to say anything to your soon-to-be ex-host. Thanks for the object lesson, GeekStorage.)

And back to Ubiquity ... I do work in IT as a linux/network systems administrator. We don't have any virtual environments yet, let alone Virtuozzo, so no, I don't have any suggestions on how Ubiquity could do a better job, Corey. I do know that the Dell MD1000 15x 1TB RAID5 arrays I work with on a regular basis can rebuild online in less than 24 hours. So, as an outsider, yeah, it's a little hard to understand why it's taken a week to rebuild less than half of the RAID array, but I've seen threads on here where it was the same story with other VPS hosts in terms of how long it took the rebuild a RAID. So, maybe I just need to accept that a VPS solution can be basically offline for a few weeks while a RAID rebuilds. Definitely a great selling point on reliability.

I just know that when Ubiquity repeatedly tries to tell me that my server is not really unusable, I infer that either:

1) You don't know what's really going on on the customer side (I did offer to provide Ubiquity with a login to the server with sudo access so you could see for yourself, but that was another thing I never heard back from Ubiquity about).

-or-

2) You don't care.

When mail isn't being delivered because sendmail is constantly timing out, when IMAP mailboxes cannot be read from/written to before the connections time out, and when Apache connections are routinely timing out, the server is UNUSABLE. Even if it's technically still "online" and can be logged into. It's not like I was dying to waste a good chunk of a day migrating my server to a new host.

I'm sorry to be harsh, but I'd be less angry if I didn't have you guys arguing about whether or not the server is really unusable and just cop to the fact that yeah, while the RAID rebuilds, the performance will be really poor.

Luckily, I rsync my data to another server at home. When I migrated to my new host, I had to use the home backup for virtually all of the recovery to the new host because data access on the Ubiquity VPS was so slow -- it was faster to upload it from a 768 Kbps (upstream) home DSL connection! It's taken me the better part of 2 days to delete my data on the old Ubiquity host because the I/O performance is so poor.

[qwidjib0]

I can't speak for any of our other staff BTamlan, as I honestly don't know who you are in our system - but I have no intent to argue with anything you have to say or suggest that you're not an intelligent person, my intent was instead to see what we may have missed and try to work with you.

Right now load on vz009 hardware node is at 20.0 due to the IO problems (was kept below 2.0 before the RAID restore).. it's a dual quad-core Xeon server and all VPS I've observed on the node have been perfectly functional, websites still load at a reasonable pace (example: http://1338.vz009.ubiquityservers.com/), emails are being sent - that's how I came to my conclusions and I assume how are staff would have as well. I understand it's not optimal, and if you've concluded that it's unusable - I have nothing to gain by disagreeing with you. Unfortunately the slowness was a necessary substitute for losing all of your data and having a day of complete downtime before we configured a new node from scratch, in what I can see, and if you have something we could do better that would not result in a negative public review as if this was just how our service functioned - I would have interest in discussing those ideas with you in private.