How much should you expect to pay for a dual dedicated server setup?

Associate
Joined
28 Jan 2007
Posts
1,586
Location
London
Wasn't sure where this belonged so I'm gonna brave GD :0

I'm looking for a new hosting provider for my workplaces ecommerce store and judging by our traffic and database sizes we have decided to get a dedicated server.

I've looked at Rackspace, hearing they're the best in the field, and they're charging £800 a month (based on 2 year contract) for the following;

  • 1 x Dedicated Cisco ASA 5505 Firewall
  • Web - Dell 2950
    RedHat ES 5
    Single Quad Core Xeon
    4GB RAM
    2 X 73GB SAS Hard Drive
    Unmetered Managed Backup

  • DB - Dell 2950
    RedHat ES 5
    Single Quad Core Xeon
    4GB RAM
    5 X 73GB SAS Hard Drive
    Unmetered Managed Backup + MySQL Agent

  • 24/7/365 Fanatical Support
  • 100% Network Uptime
  • 1hr Free Hardware Replacement
  • Rackwatch Platinum
  • Set up Fee – FREE during December
Hosting/servers etc isn't my expertise so I would appreciate any advice from those more experienced.. is this a good solution and is the price fair? Are there any other providers worth considering?

Thanks :)
 
One of my clients with rackspace has 2 boxes, one windows and one linux with a shared firewall (with other clients) and that comes in at 890 a month so the price seems pretty fair to me :)
 
One of my clients with rackspace has 2 boxes, one windows and one linux with a shared firewall (with other clients) and that comes in at 890 a month so the price seems pretty fair to me :)
That's the the same price they initially gave to us but I got them down to £800.
That does make sense :o If a mod/admin can move this thread please.

Btw we've been put on Rackspace dedicated servers before (also with a load balancer) by our magento project managers/developers and the performance difference is immense (especially back end, saving products, etc) compared to shared hosting that we've been advised to 'move back' onto recently ;/
 
We have 2 servers with Rackspace and they are awesome to be fair....not the cheapest but if you want peace of mind its worth it.

You can usually get a further discount if you pay the full contract up front. From memory we got about an extra 12% off doing that - I cant recall the exact specs as they were taken out at different time, but we got one server at 345 / month then the latest one worked out at £295 / month (but obviously had to pay £7080 in one hit)
 
That hardware is a little out of date. I think the PE2950 is based on Penryn rather than a R710 being based on Nehalem (where the memory controller is moved to the processor dye).

They are expensive. I think I worked out we're paying 5 or 6 k for server hardware costing 1k or less. I know there are many overheads on top of that, of course. I've used their support a few times and it's been second to none. Also, their customer portal is the best I've seen so far.

The only other option I've considered was Zen Internet.
 
Their customer support has been top notch so far, I started out contacting them via their live chat on the site and got most of my answers from there, they then called me and explaining further. Most pleasant business I've dealt with in a long time tbh.

I think we're gonna go ahead with them :)
 
A company that can afford to cover themselves if it does fall over.

Rackspace support is second to none in my experience.

I can garantee 100% that a 100% up time sla will be broken..... I bet in the small print it just says you will be refunded for the time its down so about £1 an hour...

unless possibly that its a distributed system such as google where if one data center is nuked from orbit you will jsut connect to another...
 
Yeah, you're buying the service more than anything, but it really is excellent. Whenever we have raised a ticket for something to be installed or changed on the server its always done within about 20 or 30 minutes.
 
What kind of company claims that?

A lying company. In essence, what this means is that they're willing to lie to you about their uptime and compensate you for downtime, whilst charging you above the odds for their marketing lies.

Rackspace are a big company, I could believe they have a few 9's which would be why you're paying a price premium, but considering that blatent lie, I'd probably head else where based on principle alone (though to be fair that'll be down to marketing/managers and probably not the techs).

On top of that, it doesn't mention power uptime, which many people learn the hard way. Its just swell having network when some data centre monkey has tripped on the power bar where all your servers are racked, take 30 minutes to plug it all back in, and you find out this isn't even covered by their SLA.

Makes it hard to know what you're paying for, and if its worth it. They'll probably be decent though, as far as I know they have a good reputation. Its possible that the market they're aiming at doesn't have enough clue to know that they should be complaining, or it might just mean they're good.

100% is a lie though. Do not believe it.
 
My company has a 100% network/power SLA with our provider and we've never claimed on it. The DC had a partial power outage at one point but dual PSUs and a redundant network meant we didn't notice one jot.

£800/month isn't a bad price (especially for Rackspace ;))

Few points to consider: (deep breath)

Why only one firewall? (single point of failure) Are you sure you need two separate servers? Why so much storage on the db server? RAID 1 for the web server (where's the hot spare?) and RAID 10 with a hot spare for the db server I presume; if they're proposing 5 disks in RAID 5 that's a no; 4x RAID5 + 1 hot spare is also a no as RAID 5 is horrible for db performance. Why not 8Gb RAM in each server? RAM is cheap. And do you have dual power supplies?

And why 2950s, they're old models although that's being picky. Personally I love R610s and R710s.

I would also look into the exact details of the backup system - how frequent are backups made and how long are they retained for? Are they kept offsite, can you download them at any time?

And do they monitor the servers? How frequently? One company (not this one) adertises that they check every 15 minutes to "prevent false positives"....monitoring should be every minute or more.

I would post on www.webhostchat.co.uk to get a range of quotes from excellent companies :)
 
I can garantee 100% that a 100% up time sla will be broken..... I bet in the small print it just says you will be refunded for the time its down so about £1 an hour...

unless possibly that its a distributed system such as google where if one data center is nuked from orbit you will jsut connect to another...

Right, but then your app would need to be distributed like say gmail, which manages to have downtime alibeit they usually say this is due to pushing out broken software. :P
 
Their customer support has been top notch so far, I started out contacting them via their live chat on the site and got most of my answers from there, they then called me and explaining further. Most pleasant business I've dealt with in a long time tbh.

I think we're gonna go ahead with them :)
Um you're talking to a sales team, they're hardly going to be rude ;)

Shop around first, not so much to save a couple hundred quid a month or so but to ensure you know more about what you're buying etc. See my last post :)
 
My company has a 100% network/power SLA with our provider and we've never claimed on it. The DC had a partial power outage at one point but dual PSUs and a redundant network meant we didn't notice one jot.

Its entirely possible that this is the case, but unless your monitoring it yourself (which I'm sure you are) you can't honestly believe they're going to tell you when they have a 30 second blip.

The previous ISP I worked for, I had a lot more exposure to what was going on in ops. Although I don't suppose they all work the same, the situation was pretty simple.

Nagios checks a server every 60 seconds, after 5 minutes of not responding it is marked as down. Hosting with them, this would mean that it is possible your server has been unaccessible for 4minutes and 59 seconds, which with no other outages (not in my experience) gives you 99.95% availability, without ever actually been marked as down, meaning you mark it at 100%.

Its easier to sell the above scenario at 100% and expect nobody to notice, than to do so at a level of honesty (and better monitoring) where you provide the same service but lose non-technical customers to the 100% guy.

But when you're talking about extreme high availability, http://en.wikipedia.org/wiki/High_availability, do you really believe you'd notice 31 seconds of downtime a year? You'd probably be paying quite a bit more for 99.9999% than the fake 100%.
 
Assuming you're hosting with any reputable host, network availability is nowhere near as important as the time it takes someone to fix a server that's had a major hardware failure,or the day to day needs of installing, updating and generally managing stuff. If I want to deploy new functionality to a site its nice I don't have to wait 24 hours for someone to respond to a ticket.

Rackspace sell on their service, and they do it very well.
 
Its entirely possible that this is the case, but unless your monitoring it yourself (which I'm sure you are) you can't honestly believe they're going to tell you when they have a 30 second blip.
The company I run is a hosting company so I'm pretty confident :)

Nagios checks a server every 60 seconds, after 5 minutes of not responding it is marked as down. Hosting with them, this would mean that it is possible your server has been unaccessible for 4minutes and 59 seconds, which with no other outages (not in my experience) gives you 99.95% availability, without ever actually been marked as down, meaning you mark it at 100%.
Agreed - like the company I mentioned earlier who monitor every 15min...very odd. Mind you their site could be outdated I suppose.

I've also seen companies operating out of Tier II datacentres and claiming impossible levels of reliability, and companies which have told off clients of theirs on Twitter for mentioning that they were having power problems!

At the other end of the scale we have new clients calling to migrate over because they were starting to think that 48 hour "maintenance windows" with their current one-man-band host were a little unusual...mad world.

Edit: And yes network issues are a tiny item compared to service and so on - it's no good having an ultra-reliable datacentre if nobody can help with your server issues.
 
Back
Top Bottom