Virtual server farm goes titsup TWICE in a month UK hosting firm Fasthosts has suffered its second outage in a couple of weeks as its virtual private servers went down on Sunday afternoon. According to Reg readers, the VPS service had already gone offline on October 31 for around 30 hours and has now been down since yesterday afternoon. Fasthosts system issues page on their …
You want good service levels. You deserve them. You are the customer.
Every minute this stuff is down will cost your business money. Your own customers will drift away and find another supplier.
Thinking about it, your hosting provider is serioulsy important. Your business could easily fail with downtime of this magnitude.
So why did you go and buy the cheapest pile of shite hosting service which for years has had a truly awful reputation for reliability, tech support and customer service? Eh ... answer me that one?
competitive tendering?
beancounters?
our boss plays golf with their boss?
lots of possibilities...
IME, for instance, beancounters and lawyers love contracts with penalties clauses and so on. These things are of course no help at all with getting the service back, but that's not their problem...
People still using Fasthosts? With their track record? Do they not remember all the previous outages? Or the time all the passwords were stolen?
As someone who was a customer for over 5 years, when I finally jumped ship in 2010 their service and reliability were at an all time low.
If you lose money because of an outage caused by Fasthosts you have nobody to blame but yourself.
I can only agree with Dr Who and AC above, firstly you get what you pay for and secondly do a search before committing.
fastshosts have a reputation and it ain't generally good.
It's simple economics, the cheaper you sell them the less money you have to invest in staff and infrastructure.
If you are that reliant on your website for your business consider investing some of the profits and decent hosting and independent monitoring or at least have a business continuity plan.
Cheap generally works but when it goes wrong you'll be offline for hours or days....
Cheap services are always cheap for a reason. Even allowing for economies of scale, cheap services are usually cheap because corners have been cut.
The use of inferior technology, lack of resilience, slow recovery technology - these are places where you can cut corners to keep the price down.
So don't be surprised when cheap stuff fails more often, stops completely when it fails and takes longer to recover when fixed.
I'm not in the UK so have never really heard of fasthosts but looked them up and looks like one of those cheap ass hosting providers.
Those who use them and get mad when there is downtime have nobody to blame but themselves.
Here I am paying US $100/mo to co-lo my 1U server for *personal* email/web hosting, I consider that cheap too.
stupid customers, fasthosts is for kids!
People still host with Fasthosts. Their domain registration is okay - no better or worse than many others, and not too bad on pricing - but their hosting services? When it comes to hosting, after their handling of the Usmanov affair a few years back, I wouldn't touch them with a 10m Cat6 cable.
anyone with inside information know what storage platform(s) they are running and what the interconnects look like? I look after a 400+ VM VMware farm, and from what we've seen in the past, it's usually flagging shared storage that causes the Linux file systems to switch into read only mode whereas the windows VMs by default appear to be more resilient to back -end storage drops. My bet is that there are unknown storage related issues causing the problem. ZFS perhaps?
As someone else commented, underlying storage problems will cause Linux to force remount the filesystem read only while windows will happily continue using it in read/write mode...
But this is not a case of windows being more resilient, it is of linux trying to avoid causing data corruption. Depending on the root cause of the storage issues, you may get lost/corrupted writes and the linux approach is a failsafe (dont write anything) while the windows approach can and in my experience often has resulted in severe corruption of data.
Of course it entirely depends on what exactly is causing the storage problems, but the guest os running inside of a vm has no way to know.
SLAs are pretty much always like this. We'll give you a tiny amount of money back, and cap it at a slightly less tiny amount of money, should it go completely pear shaped and the 5 nines run off into the distance screaming. In this case, if you notified Fasthosts that your service was down, you'll get a whole month service credit. Wow. I've seen much bigger services that offer pretty much the same - here, massive company, we'll give you a day's service credit in return for your day of lost trading.
Although I do agree in this instance that expecting events of this nature never happening on such a cheap service is foolish. If your service was business critical you implemented a standby on another provider didn't you?
That said, like "unlimited" broadband, the practice should be stopped. 99.99% guarantees should only be made by those that guarantee it, not with a big asterisk next to it linking to a passage offering something quite different. In this case 99.99% uptime will have been added to Fasthosts webpage because, like unlimited broadband, that's what everyone else was offering, and they were losing customers by not doing so.
We used fasthosts VPS not because of price but because they have a snapshot system that backs up the servers every hour, in disaster recovery you can recover the entire server quickly if there is a major problem.
We use Windows servers and with their Mircrosoft endorsement it all seemed like a no brainer.
What we didn't expect was this kind of major infrastructure fault. I'm shocked how Fasthosts could get this so wrong.
We are having a nightmare at them moment because we have multiple VPS servers with them.
a quick google brought me to these;
http://www1.euro.dell.com/content/topics/global.aspx/casestudies/en/emea/eu/fy2011_q2_id1670?c=za&l=en&cs=RC1078563
http://www.microsoft.com/casestudies/Windows-Server-2008-R2-Datacenter/Fasthosts/Hoster-Expands-Business-Cuts-Costs-by-Using-Hyper-V-to-Offer-Virtual-Servers/4000007585
"..Working with Dell and Microsoft, Fasthosts deployed a cloud infrastructure with Hyper-V , Dell PowerEdge servers and Dell EqualLogic storage."
"Fasthosts uses Dell PowerEdge M610 blade servers and Dell EqualLogic PS6500 storage area networks (SANs) to host its DCoD infrastructure."
so.. now we have the parts list, anyone likely to guess what firmware versions the storage is running on and if there have been some recent changes? 5.1.1-H2 or something else? Looks like the H1 release could bring down the storage during an upgrade from a quick Google..
All speculation of course until fasthosts can give a clear indication of what has gone wrong!
Still no resolution to the virtual hosting issues. Even now, they have at least 150 virtual servers affected (which they admit to), but I'm guessing the actual number is a lot higher. No sign of any resolution in sight - it's a complete meltdown.
My server has been up and down 7 times since last weekend, and not responsive for > 80% of the time, each time they restart it, but they can't actually fix the issue which is causing this for everyone, and can't or won't give an eta for a fix. I've already lost all my reselling customers.
Their website is still accepting new virtual host customers, and claiming there are no 'system issues' ongoing - how can this be the case given their virtual server platform can't run a server for more than a few hours without a reboot? I think the legality of that is dubious, to say the least.
They have now posted a fix on their site which at least sets things up so we can in theory restart a server ourselves - so isn't that nice for them. They don't even have to get any more support tickets, they can just sit back, while we just have keep pinging our servers and restarting them every couple of hours when they go down.
The Register 
Biting the hand that feeds IT
Copyright. All rights reserved © 1998–2024
