Register1 Down?
Discussion
Spookily like the problem I'm having with Easily (see other thread).
I wonder if they are connected?
>> Edited by arcturus on Tuesday 1st March 12:51
I wonder if they are connected?
>> Edited by arcturus on Tuesday 1st March 12:51
The Register said:
In an email to customers, the company said the power failure was caused by the "incoming mains into the UPS feeder board blowing up causing the power to go down".
Surely an email which they wouldn't get if the power was down and emails not working.
Easily email still down too.
arcturus said:
Is there anywhere we can get more info from or maybe an ETA for the repair, 'cos me (and several of my customers) are getting p**d off with no email.
My servers with Serverstream were returned to normal in less than an hour of the outage. I gather the fire suppression systems kicked in too.
This kind of outage is utterly unacceptable. it should be impossible and there is no excuse for it happening once let alone several times.
If anyone needs to move as a result of it do let me know....I don't want to poach any business from the other hosters on PH so if you are with Jamie or Rico or Muncher I can't help on this one (although I'd love to help them
, but if this outage is one too many or if anyone is looking at preventing this sort of thing happening again do let me know. I am very happy to do a deal for fellow PHers. We don't do shared hosting as the likes of Jamie do it too well for me to enter that market, but if there is anything else I can help with email me via my profile and I'll do whatever I can.
Cheers
Dan
If anyone needs to move as a result of it do let me know....I don't want to poach any business from the other hosters on PH so if you are with Jamie or Rico or Muncher I can't help on this one (although I'd love to help them
, but if this outage is one too many or if anyone is looking at preventing this sort of thing happening again do let me know. I am very happy to do a deal for fellow PHers. We don't do shared hosting as the likes of Jamie do it too well for me to enter that market, but if there is anything else I can help with email me via my profile and I'll do whatever I can. Cheers
Dan
I think what this event shows is that no setup is 100% fullproof reliable. Many are 99.9999% but there is always a chance a problem will occur.
A huge number of big multinational companies host in RedBus... they're not stupid. They host there because it is one of the best, if not the best place in Britain to host.
You'll be getting all the other non-Redbus people boasting that "we can do better" without any real proof to back it up.
Thats not a go at Dans... give more information and PROVE why you're better than Redbus and maybe some of us will move over. I know that I'm chuffed to bits with the service i'm currently getting.
C'est la vie.
A huge number of big multinational companies host in RedBus... they're not stupid. They host there because it is one of the best, if not the best place in Britain to host.
You'll be getting all the other non-Redbus people boasting that "we can do better" without any real proof to back it up.
Thats not a go at Dans... give more information and PROVE why you're better than Redbus and maybe some of us will move over. I know that I'm chuffed to bits with the service i'm currently getting.
C'est la vie.
arcturus said:
Dan, who do you represent, there is nothing on your profile that gives a clue.
Do you have a website we can look at? Why are you different to the others? Couldn't this happen to you?
Dan runs Carrenza which hosts PH. I'll let Dan explain the architecture of the bunker we're in but it's something to do with a completely redundant back to back twin independent building config type thing. He did explain it to me once and it made a lot of sense
The best answer I can give you is the straight one Rico. The buildings are all the same. the way they are filled and managed are not. in any N+1 environment it is crucial that the N+1 balance is maintained across the building that means hard truths about how much kit can go into each rack and tight control of the power going to every feed.
most of the facilities in the UK were designed over 4 years ago - back when servers had power supplies that consumed half or less of the power of todays devices and when most of them were assumed to be bigger then the 1U or blade format devices that are common today. The highest capacity facility in the UK in terms of power per SqM is Level3 in Islington. they can handle 1 kW per SqM that means about 2kW per rack in a normal IDC. Now look at the specs of most servers and you'll see that even a 1U entry level device has a 350-400 Watt supply and will be using at least half of that. that means 10 1U servers in a rack in the highest capacity IDC in the country is the most the N+1 can stand. More than that and you are eating into any redundancy.
in recent years the industry has been desperate for clients and they have been happy to take the money off anyone who'll pay and said they can provide extra power to racks and additional cooling which in a true N+1 environment is impossible. Now the industry is hotting up and IDCs are filling up and every so often the usage is more then the infrastructure can cope with. I am categorically not saying that is what has occurred here as I have no knowledge barring the articles on El Reg to go on, but this is a known phenomenon and it is hurting.
I don't know if I would claim to be better, but I know we don't compromise on system integrity to win a deal (I have walked away from over £1M in business this year where people had unrealistic power requirements) We keep our racks emptier, clients have power related to the floorspace they buy from us thus ensuring it is related to the N+1 loading on the facility and if they want a rack full of blades it is possible, but it will be in the middle of a big empty space to allow for the power. It might not be the cheapest solution, but it will work and it will have an utterly uncompromised N+1 setup. guaranteed, and not just by marketing speak but by charts and careful use of building management to ensure that Power and cooling SLAs are met.
If you want to know more let me know.
most of the facilities in the UK were designed over 4 years ago - back when servers had power supplies that consumed half or less of the power of todays devices and when most of them were assumed to be bigger then the 1U or blade format devices that are common today. The highest capacity facility in the UK in terms of power per SqM is Level3 in Islington. they can handle 1 kW per SqM that means about 2kW per rack in a normal IDC. Now look at the specs of most servers and you'll see that even a 1U entry level device has a 350-400 Watt supply and will be using at least half of that. that means 10 1U servers in a rack in the highest capacity IDC in the country is the most the N+1 can stand. More than that and you are eating into any redundancy.
in recent years the industry has been desperate for clients and they have been happy to take the money off anyone who'll pay and said they can provide extra power to racks and additional cooling which in a true N+1 environment is impossible. Now the industry is hotting up and IDCs are filling up and every so often the usage is more then the infrastructure can cope with. I am categorically not saying that is what has occurred here as I have no knowledge barring the articles on El Reg to go on, but this is a known phenomenon and it is hurting.
I don't know if I would claim to be better, but I know we don't compromise on system integrity to win a deal (I have walked away from over £1M in business this year where people had unrealistic power requirements) We keep our racks emptier, clients have power related to the floorspace they buy from us thus ensuring it is related to the N+1 loading on the facility and if they want a rack full of blades it is possible, but it will be in the middle of a big empty space to allow for the power. It might not be the cheapest solution, but it will work and it will have an utterly uncompromised N+1 setup. guaranteed, and not just by marketing speak but by charts and careful use of building management to ensure that Power and cooling SLAs are met.
If you want to know more let me know.
rico said:
Interesting stuff. So basically, if a cabinet has more than 10U of servers in place, it cannot run to its full specification in the UK?
It will be interesting to hear Jamie's viewpoint on both 1) what happened today and 2) what you've said.
If a rack is trying to draw more then the per Sq M equivalent of 1kW/ sqM of IDC space then yes it is blowing the N+1 - not just in the UK but in pretty muck all IDCs globally...
rico said:
A huge number of big multinational companies host in RedBus... they're not stupid. They host there because it is one of the best, if not the best place in Britain to host.
I agree with what you say here, however, todays outage was the 3rd major one at Redbus that I know of (previous ones on 9 Nov 04 and 16 Jun 04) and I'm starting to get a bit fed up with it. To me it says that there is something fundamentally wrong with the setup there.
Fun Day 
Will stick a longer answer on later once I'm settled in again.
Suffice It to say that, as has been said, one of our suites in one of the DCs we are located suffered a Partial power outage today.
Our core network remained up at all times due to our own UPS systems, however all of our connecitivty providers in that site went down.
Service was resumed within 40 mins to the majority of the affected kit.
Clients with multi-homed kit in other sites were kept live as intended.
Never fun when a DC goes pop, but thats why we have multiple sites for the bigger kit...
I think I will migrate Reg1 Clients to SoV, HeX, bless it, does seem to be having more than its share of issues of recent..
Right, Back to it.
Enjoy yourselves
Will stick a longer answer on later once I'm settled in again.
Suffice It to say that, as has been said, one of our suites in one of the DCs we are located suffered a Partial power outage today.
Our core network remained up at all times due to our own UPS systems, however all of our connecitivty providers in that site went down.
Service was resumed within 40 mins to the majority of the affected kit.
Clients with multi-homed kit in other sites were kept live as intended.
Never fun when a DC goes pop, but thats why we have multiple sites for the bigger kit...
I think I will migrate Reg1 Clients to SoV, HeX, bless it, does seem to be having more than its share of issues of recent..
Right, Back to it.
Enjoy yourselves
Gassing Station | Computers, Gadgets & Stuff | Top of Page | What's New | My Stuff