After months of planning, we are finally in the homestretch towards releasing our (until now, Uber-secret) VPS.net Cloud.
So as such, time has come to start a blog, to keep everybody informed on the latest and greatest features we are working on for the VPS Cloud, as such, during the upcoming days and weeks, we will post details on the new website, the Cloud features, the user interface, and much more.
VPS.net is a new brand by that great hosting outfit, the UK2 Group, owners of 10TB, UK2.NET, Midphase, etc ..
So keep checking this page every day (or even several times a day, things are moving fast), we expect to release VPS.net LIVE on January 31st, but will be looking for beta testers in the upcoming few weeks 
Is this gonna be a kinda MidPhase meets MediaTemple? I need to migrate about 80 websites away from a couple of web.com VPS servers (I’ve been using them since they were vservers before interland bought them). I tried MediaTemple out for about 9 months during 2008 and the cluster deal didn’t quite work. Instead of; when a bit fails the rest takes over, when a bit failed it took the whole cluster down. Continuous firmware, hardware and general we can’t work out what’s going on type problems. I was planning to try UK2 and MidPhase hosting (I know it’s the same umbrella, but the US has a different approach etc etc) in 2009 before the definitive move from web.com, but I’ll have to put that on hold now and see what this turns in to.
I’ve rented domains from UK2 successfully for nearly a decade now so the precedents are good. Make sure it’s stable though, VERY stable, clients websites go down, email stops and we are left saying “it’s someone else’s fault”, never sounds too good. Walk into the middle of the cluster and pull the plug out of three servers / load balancers or both and see what the cluster does, take out some routers, unplug some network cables etc etc.
If we could have multiple SSLs, no limit MySQL DBs and domains all in one account, with tons of bandwidth and VPS type access, with remote SSH instant reboot etc etc (I’m sure you know the score), that would be the one stop shop solution. Not one of these servers and three of these add-ons and another one of those servers for the ‘other’ sites we have to host.
Good luck anyway, we’re watching and waiting…
P.S. Submit Comment doesn’t work in FireFox 3.0.4, changes to not a button but then nothing happens, bad omen? Well see…
Hi Andy
I’d be interested in hearing more of your experience, when you say “a bit failed and took the rest over” .. can you give me an example ?
Our intention is to provide a VPS “node” with fixed features (ram/cpu/hd/bandwdith), you can then grow your setup it by buying more of them (we wont want to get into the whole “buy more cpu cycles for x, and more ram for y”, Amazon does that and it’s complicated as heck .. our objective is to make it as simple as possible.
And you be happy to know we ARE indeed pulling cables, it’s the funnest way to test .. a blog post will come on that experience soon
The theory with a cloud / cluster is that resources are more evenly shared out and there is more to go round right? You are not simply sharing one physical box, you are sharing the whole cloud, the whole bunch of boxes…
What the problem was (and still appears to be http://weblog.mediatemple.net/weblog/category/system-incidents/2008-11-24-gs-grid-service-cluster2-availability/) with MediaTemple, is that when one box, one hard disk or one email server failed, instead of the rest of the cloud smoothly taking up the strain, the process of the rest of the cloud taking up the strain, took the rest of the cluster out of service as well.
I remember a specific case where a hard disk failed somewhere in Cluster.2, where my test sites where, and in theory all the data on that disk would be restored to another disk in the cluster somewhere and everything would carry on as normal. But there was a problem in the firmware somewhere apparently that meant that the process of replicating that disk took priority over other services, so the rest of the cluster basically stopped serving websites and sending e-mails whilst a disk was restored. That restore took longer than expected, and by the time it was complete the backlog of e-mail was huge and then the cluster struggled to cope with the huge demand for sending queued emails.
So instead of a hard disk failure causing 2 hours downtime on one shared server, that disk failure created havoc across the whole cluster for nearly 24 hours.
That is the risk, if there is a single failure somewhere within the cloud; that brings down the whole cloud, a small isolated problem becomes a huge generalised problem. Of course that is exactly what is NOT supposed to happen but in the case of MediaTemple it seems to still be happening.
If VPS.net can get a totally resilient cloud going then we, well at least me, will be VERY happy and a lot more relaxed as I won’t be constantly worrying about a server failure bringing down 40 websites, 40 lots of company e-mails etc etc
In the post http://vps.net/fluffy-vps-cloud/#more-16 , the key phrase for me is “…utilized (sic) to deploy VPS’s in seconds that are amazingly scalable, not bounded by a single server limitations and can be self healed in case of disaster in less than 1 minute.”
If I can have my VPS in seconds, set up my websites, email, ftp, ssh, php, mysql DBs etc, AND know that if sendmail fails on that VPS it can “self heal in less than a minute”, or if the entire VPS fails it can “self heal in a minute”, then count me in!
Hi Andy
I understand your concern, it seems today, he biggest problem with a Cloud setup is I/O .. to avoid problems we decided to limit each Cloud NAS to 10TB, this ensures that less traffic is flowing via a single Gbit NIC, but it seems tat some providers are over complicating this by trying to do “bigger is better” setups, and tie in the entire NAS as a single volume .. we wont fall for that, it creates alt of problems later on when the volumes become saturated.
On your instance, I am quite suprised with the problem you experienced, we will be using RAID 5/10 on our NAS Nodes, so if an HD fails, you wont even know it, then of course, it’s a matter of the NAS RAID rebuilding itself (no downtime there) to bring the RAID in question back to normal status.
But as you know, this is not something new .. it’s just an every day setup utilized on a larger Cloud environment .. we are not going to be re-inventing the wheel here, we don intend to compete with the Likes of Amazon for a uber-complex system hat gives you a headache while trying to order it alone, we instead are concentrating in utilizing tested and true technologies with a unified custom backend created by us to make it a Cloud setup you can trust.
Your point is correct, this is not just a group of servers, that is a Cluster, not a Cloud, if a server fails, it needs to take over another somewhere in the cloud and be back online in no time (we strive for less than 60 seconds)
Sometime next year we will even deploy additional redundancy for the Data nodes, so that in the event of a catastrophe to the Data systems (full failure of the entire NAS, highly unlikely, but hey …. ), another kicks in instantly, no need to restore from the backup NAS .. it’s a triple redundancy complex system .. allot of work still to be done on it, but in the end it will be worthy
Hi Andy,
Do you think VPS is the way of the future?