Forum Moderators: buckworks
I appreciate your taking the time to read this and any guidance received will be greatly appreciated.
I have a client who has an account with a certain webhost. The webmaster who put him on to this company has since moved on. From the first few times I looked at the client's site I noticed that it appeared to experience alot of down time. So much so I signed up with Server Monitoring UK to keep track of the site. The first 30 day period will end tomorrow and here are the stats; (In fact the site is unreachable as I write this and I have received alerts to this effect)
Monitored Since: 02/08/2005
Report Updated: 03/07/2005
Outages: 91
Down Time: 8 hours, 40 minutes
Availability: 99.319%
How does this 30 track record rate in your experience?
It seems excessive to me but I have no idea what the average is or what is considered "sub par"
This client sells very high end products and it appears that these seemingly frequent outages could be costing him alot of money.
As of now I intend to recommend he change hosts but I would not want to be responsible for his site's performance going from bad to worse.
I'm looking for a new host and I like bells and whistles as much as the next guy.
But uptime trumps all other considerations IMO.
Thanks for your time and consideration,
Ken
I've had one network outage of under half an hour on my main (dedicated) server in the past year, and it hasn't been rebooted for over 15 months. That's acceptable (probably better than average).
I have tested another company, that one was 26 outages totalling over 16 hours in one month. Hopeless.
The outages will certainly cost him money, as well as search engine rankings (quite likely to get dropped if the site can't be found after multiple attempts). It also gives an unprofessional appearance. Don't accept it, but DO test the new host first!
I can think of a few ways I might do this but just how would you do it?
Thanks for the input, I'm more convinced than ever that this host needs to get fired.
Ken
Also use this time to test the server can actually run the technology you need, has the right paths for scripts, email is easy to configure, support responds to questions, etc. Often there is a trial period or 30 day money-back guarantee available.
My feeling only gets stronger that this client could do much better than the host he is with. And it will have a direct bearing on whether my efforts to market and promote his business are effective. The best marketing in the world is rendered ineffective when prospects cannot download the site.
Thanks again,
Ken
As for acceptable downtime, most hosting companies swear by 99.9% uptime. My general policy is that if they are not maintaining that uptime, then we send an invoice to the hosting company for the amount of revenue that was lost.
Now bear in mind that this doesn't work for most hosting companies. It works for us because we spend alot of money on hosting, so it makes big bucks for the hosts.
I was looking at a host yesterday that guaranteed 100% uptime but in looking at the fine print anyone would be hard pressed to prove an outage was the fault of the host. There were so many "ifs", "ands", and "buts" that at least in my eyes, the 100% guarantee was pretty much meaningless. The most reliable and best way to evaluate a host is by getting input from parties who have nothing to gain. But asking for recommendations is a sticky wicket. If I did so in this forum it would invite all sorts of negatives which was pointed out to me when my original post asking for them was deleted.
I understand the rationale of the moderator 100% and I value being able to come here and benefit from the wisdom of the board's users so I don't want to run afoul of the rules.
What complicates matters even more is the fact that this site was originally created in Cold Fusion. I did get some recommendations from another source but the top ones use UNIX which is not Windows which is what CF requires.....I could completely redo the site and use PHP or another alternative but that greatly increases the amount of work involved....
Anyway, I'm starting to ramble so thanks again for your valuable input.
Ken
99.9% means outage allowed 8 hours, 45 minutes per year
99.5% means outage allowed 1 day, 20 hours per year
Linux and Unix systems are very stable, so I think getting a web host that offers 99.9% guarantee will not be hard to find.
Yes, and this particular host had over 8 hours down time last month.
And I don't doubt that UNIX or LINUX hosts are more stable, but from the "migration" consutants I have heard from I'm hearing that it would be best to go with another CF host so we'll see.
Thanks for the input.
Ken
In working with the business owner we have figured out he can greatly pare down his product selection by combining similar products on one page and eliminating products that are generating no sales or interest. So, actually it looks like we will be rebuilding the site minus Cold Fusion. This should open up alot of hosting possibilities.
I have gotten some suggestions so far but if anyone has any more (Unix, Linux hosts), please respond or private message me.
Thanks again,
Ken
[edited by: Kenzilla at 2:34 pm (utc) on Mar. 15, 2005]
It may not be the host at all, it may be some flaky router along the way to the server.
Best to do a little homework before you make a lot of extra work for yourself.
Yes.
Ken
I have done this successfully with 8 completely independent providers (well almost completely).
Find a DNS provider who is not with either of the hosting providers.
Provide them with DNS info for the first host IP/name then provide a second set of IPs with same name. The provider monitors the primary host, and if it goes down, automagically provides the new set of IPs to client name resolution requests, ergo "redirecting" the visitors to the backup site.
There are several requirements and benefits to this solution.
You or the DNS provider (actually they were called edge service providers when I implemented it, or whatever the cool word-of-the-second is now [ why do tech wanna-be gurus come up with alternate names for existing solutions then repackage them as a new thingamajig?]) will have to make sure you synchronize the backup server(s) with the primary server.
In my client case, the redirection happened not only when the primary server was down but all the Time. Why? Because in case you are in China, and wanted a fast response why hop across the ocean when you could just visit a backup server in Hong Kong? The edge service provider, looking at the client's IP could determine which server was closest hop-wise (not geographically) to them. This provided redundancy and increased speed.
Of course it cost an arm & a leg, and the sync process was seriously critical, but I could get sub 10ms anywhere on the planet.
So, try to find a provider that will give you the alternate or fallback DNS solution. This way, your client will be happy as a bug.
In one implementation the fallback was the client's own server in-house, off of a DSL line. Yes, it sucked mud, but was still better then a 404 Not Found! Plus that server was always up to date, and did not require a separate cost for ISP...
