Web Hosting Talk : Web Hosting Main Forums : Web Hosting : Any 100% uptime hosts (redundant servers) with ASP and support Persits components?

[Ron R]

Hey guys, I have an immediate need for a web host that meets the following criteria:

* 100% up-time guarantee, redundant servers or "cloud" hosted. Site cannot go down.
* Windows server with Classic ASP
* Dedicated SSL (I have the certificate already, cannot be shared. The site runs in SSL full time)
* Must support Persits ASPEmail, ASPUpload, and ASPJpeg (latest versions).

The reason the site needs to have 100% up-time is because it's a content provider that a lot of other sites link to using a single line of Javascript - if the server is down, THEIR web pages hang due to the way browsers handle JS. So even a 5 minute outage is unacceptable, as it would cause the website of all of our customers referencing the app on their sites (usually their home pages) to hang for that period. I know this can be done, I just don't know who offers it reliably and don't see the need in going the dedicated server route for a single website.

The site is also full time SSL, if a user visits in http one of the first lines in the code redirects the user to the https version of the page, so it needs to be able to use my certificate and not a shared one. That certificate would have to be installed on each of the servers the site is replicated to.

I just need some direction on how to best handle this and who to go with. I currently run my own servers, but after a mere seven minutes of down-time last week I had a lot of irate customers, so this site needs to be moved out.

Thanks in advance. (I would put a smiley here but I guess I'm not allowed until I have made five posts, so just pretend there is one here).

[NoSupportLinuxHostin]

First of all, 100% uptime is not possible. You can get really close to 100% if you spend enough money, but you will never be able to actually hit 100%. Providers that advertise 100% uptime use a service level agreement to provide the promise of 100%, but there are plenty of exceptions in those documents. For example, taking services offline to reboot a box would often times not be considered downtime as far as the SLA goes. So each time the servers needed to be rebooted, that would count as zero downtime, even though the sites would be offline during the reboot.

What kind of budget do you have for this? There are ways to deliver 99.99% to 99.999% uptime using a cloud infrastructure with clustered guests in addition to clustered hosts. Can you pay thousands per month for that level of service or are you looking to pay $5 per month? I ask about the budget because it seems like most guys who look for 100% uptime are also looking to do it really cheap.

[Ron R]

100% uptime is possible in a clustered environment, I just don't know who offers it reliably. If a server is rebooted, the load balancer delivers the traffic to the other one. That's the type of host I'm looking for.

I don't need thousands per month for this solution, I could pay a few hundred a month and run more fiber to my building here from a different ISP and accomplish the same thing. Just trying to avoid going to all of that effort for a single website. I suppose its probably going to be advertised as 99.999% uptime for legal reasons, which is fine, but you get the idea of what I'm looking for. I would be fine with something around $100 a month or so, it's not a very resource-intensive site, just needs to be up.

[NoSupportLinuxHostin]

Ron: Even in a clustered environment, there is a brief period of downtime while the cluster service cuts services over to another physical node. That is why 100% is impossible even with clustering. The way clustering works, all clustered resources (network, storage, etc) must be taken completely offline on the current node before they are brought online on the next node. Depending on the number and type of clustered resources, the failover between nodes may take several seconds or even a minute. No matter how fast the failover is, it still prevents a true 100% level uptime.

I don't know anybody who could deliver five nines of uptime for $100 per month. You might find hosts who can deliver five nines for that price point most months, but not every month. Your situation is even more challenging since it requires Windows. The licensing costs for Windows Enterprise (to get clustering support) for two physical nodes will eat into your budget.

If you already have a clustered server solution in your building, then running extra fiber might be a solution for you, especially if the recent downtime you had was related to network availability. Another option would be to colocated some clustered servers in a datacenter, since most quality datacenters are already able to deliver five nines of uptime for network services.