Remote Datacenter Wordpress Failover Solution

thewpfan · #1 07-13-2009, 03:33 AM

I am trying to setup a wordpress blog network with 99.9% uptime. My experience with hosting companies is that there are hardware and network failures that can cause significant amount of downtime. So I was thinking of creating a remote failover solution. Is it possible?

I have been searching for a remote (vps in Japan and US) datacenter wordpress (application and mysql database) failover solution, but cannot find anything on the web. Ideally, the solution should be syncing both datacenters at the same time. If server A goes down, then server B will startup. Once server A is up again, server B will send its updates to server A and server A will resume. Since this will probably taxing on server A, server B might have to be 4-8 hours delayed. It is not ideal, but I have not found a solution to keep them in sync constantly.

I have searched under failover, clustering, replication, mirroring, syncing, and high availability over search engines and forums, but have not found a robust solution. I have read about dns failovers, but many have recommended that it is not a good solution.

If you have experience with a remote datacenter wordpress failover solution, please share.

Thank you.

falconinternet · #2 07-13-2009, 07:54 AM

For one site like this I would rsync the files every 5 or so minutes. You could also use LFTP to mirror the files if you don't have that type of access.

For the MYSQL data, if the database is small you could just dump the data every 3-4 hours and then copy it over to the server. Or, you can run a MySQL slave (if you have root access to both servers). I've always had trouble running MySQL slaves, but you may have better luck.

Keep your DNS hosted at a 3rd party and set the TTL values of your A records to a low number like 300 (5 minutes) and then switch if there's a problem.

thewpfan · #3 07-13-2009, 12:27 PM

@FalconInternet
Is it possible to setup a MySQL master & slave for remote datacenters? Wouldn't latency affect synchronization?

In regards to DNS, were you thinking of a dns failover system like dnsmadeeasy?

falconinternet · #4 07-13-2009, 12:38 PM

#1: you can do a slave setup between 2 data centers. It's pretty fast. Unless you have just 1000s of insert queries, it shouldn't be an issue.

#2 DNS, yes, there's several DNS services like that one. Or you can do it manually (which I prefer), because you don't want to to revert back to the master if it comes back up, it doesn't sync the data both ways.

Another option is to update 2 databases at one time, the remote and the slave although that would require a lot of word-press customization.

MikeDVB · #5 07-13-2009, 12:39 PM

You would really need to have some way to replicate the database information - replicating the files alone won't do you any good with a WordPress installation.

If it is that important to you that you keep your blog online you may want to look into failover web hosting otherwise going with a reputable and trusted provider you should be able to get 99.9% uptime in most cases.

thewpfan · #6 07-13-2009, 01:23 PM

@FalconInternet
Thank you

@MikeDVB
I have actually searched for cross datacenter failover VPS solutions and there are few. I am currently communicating with unixy.net. I will update everybody on WHT if I go with their services. I have been with some reputable providers, but all of them have had hardware problems, network downtime, upgrades downtime, etc. Am I the only one? I would like to emulate 99.9% uptime and I think a cross datacenter solution might be the best. I am setting up a blog network for professionals. I would like to be professional too by providing a service that is redundant.

dotHostel · #7 07-13-2009, 01:26 PM

When you are used to shared hosting after some time one might believe frequent downtime is a fact of live in the hosting industry. For some extent this may also be true when you buy a VPS from inexperienced providers as you don't have full control and is subject to unscheduled server reboots and others service outaqes. The fact is, if you have a VPS or dedicated server with a top-notch provider hardly you have frequent outages -- your server uptime is likely to be counted in years and if uptime is critical you need a disaster recovery plan instead a DNS failover.

I think for your case you could deploy a couple of front-end cheap VPS (US,JP) running Squid as reverse proxy, or Nginx forwarding dynamic queries, and back-end server(s) running WP and the DBMS with frequent off-site backups. You could use DNS failover (as DNSMadeEasy) for the front-end servers.

Note: 100% uptime may be achievable "inside" the data center but rarely in the real world. If network providers who own the last-mile can't ensure quality bandwidth to their customers, how can a service (including CDNs) that is relying on these networks do this?

MikeDVB · #8 07-13-2009, 01:27 PM

I'm not against what you're trying to do, I'm just saying to to do it *right* is not going to be cheap.

dotHostel · #9 07-13-2009, 01:33 PM

Quote:

Originally Posted by MikeDVB

I'm not against what you're trying to do, I'm just saying to to do it *right* is not going to be cheap.

Agreed.

cartika-andrew · #10 07-13-2009, 01:50 PM

Quote:

Originally Posted by thewpfan

I am trying to setup a wordpress blog network with 99.9% uptime.

Just find a provider that can actually deliver 99.9%.. honestly, over the long term, this is a very attainable metric, even for shared hosting ...

How much traffic are you actually pushing a month?

cartika-andrew · #11 07-13-2009, 01:52 PM

Quote:

Originally Posted by MikeDVB

I'm not against what you're trying to do, I'm just saying to to do it *right* is not going to be cheap.

exactly Mike..

to the OP - if you want to setup the type of HA setup you are talking about, it should be because your uptime objective is significantly higher then 99.9% - and as Mike said, you should have a significant budget to accomplish this. If you are just looking for 99.9% - a good provider can meet that. Sure,there will be times with network upgrades or problems that may cause you to miss that objective occassionally, but, having a 99.9%+ uptime average over years of hosting is certainly not unheard of..

MikeDVB · #12 07-13-2009, 01:54 PM

Quote:

Originally Posted by cartika-andrew

exactly Mike..

to the OP - if you want to setup the type of HA setup you are talking about, it should be because your uptime objective is significantly higher then 99.9% - and as Mike said, you should have a significant budget to accomplish this. If you are just looking for 99.9% - a good provider can meet that. Sure,there will be times with network upgrades or problems that may cause you to miss that objective occassionally, but, having a 99.9%+ uptime average over years of hosting is certainly not unheard of..

This is definitely what I was trying to say above I just didn't use quite as many words.

If you go with a reliable provider you shouldn't have a problem with 99.9% - if you really want failover/highly available you're going to end up paying significantly for it.

Jacob Wall · #13 07-13-2009, 01:57 PM

Quote:

Originally Posted by cartika-andrew

Just find a provider that can actually deliver 99.9%.. honestly, over the long term, this is a very attainable metric, even for shared hosting ...

How much traffic are you actually pushing a month?

edit: I fail at math