Results 1 to 7 of 7
-
01-31-2017, 07:00 AM #1New Member
- Join Date
- Dec 2011
- Posts
- 3
What would you recommended as server redundancy scheme?
Hello,
we're providing shared hosting for a few our clients and our current server and OS (CentOS 4) + software (Apache and SSL) are outdated (and server itself is quite expensive at Coreix).
Now we are searching for better and cheaper dedicated server provider (with server located in UK). At the same time we are looking for a right way our clients were protected from downtime and we want to have a good Server Redundancy.
Should be a scheme with 2 mirrored servers (rsync?) + DNS failover + additional Network-attached storage (for user accounts backups) good enough?
Or there are better ways? I've read about virtualization and couldn't understand how can I use that if I have dedicated server. Should we create that ourselves or server provider can do that? And is the vistualization good enough for server redundancy task? Can someone suggest the good and cheap way for making our server protected from downtime if something happens with hardware and it becomes unreachable (so in this case other server becomes active and all is running until master server is recovered)?
At the current server we have DirectAdmin lifetime license and we want to move that to new server along with client accounts (which is quite easy). Our clients use email(imap/pop), ftp, http and usual services which accompany websites
-
01-31-2017, 09:07 AM #2Junior Guru
- Join Date
- Oct 2016
- Posts
- 180
Centos 4 is too old OS . Right now latest one is centos 7
For High Availability without any downtime you will need to setup load balanced setup where you will need 3 server one for load balancer and other 2 for services.
These data can be synced using rsync from master to set
To load balance you can use http://www.haproxy.org/
The load balancer server can also be a 4 gb vps need not be dedicated server as it uses less resources and its function would only be to redirect traffic.
Virtualization would depend on the data center or hosting company you get server from
. Some do it and some dont .
if there are more account I would suggest you stick to dedicated instead of virtualization as performance in dedicated would always be better compared to virtualized system
-
01-31-2017, 02:31 PM #3New Member
- Join Date
- Dec 2011
- Posts
- 3
Thanks for the response!
With this being a shared hosting server using DirectAdmin with around 100 clients on it I wonder if there is some off-the-shelf tool/software we can use. *What are the best methods people use? How do hosting companies do this normally? i.e. strip accounts across servers for redundancy.
It would be same data centre, but we don’t know where to start. With the large number of clients we don’t want to set up rsync and mysql replication for each client*
Any suggestions welcome
-
01-31-2017, 04:08 PM #4Web Hosting Master
- Join Date
- Mar 2005
- Location
- Labrador, Canada
- Posts
- 988
You can strive for a high-availability setup, though it is neither simple nor 'cheap' (your word).
First I suggest focusing on your backup system so that you:
- Perform a regular off-site backup (to an external location),
- Are able to restore that backup to a new location.
With that you can bring sites up at will (from backup, after a catastrophic failure). The time to do so depends on DNS TTL and data transfer speed (time it takes to copy data from your backup).
Once you've nailed that down look to a 'high-availability' solution. It's more complex, and more-complex solutions have more complex problems.
But you have an off-site backup and restore procedure to fall back on...
-
01-31-2017, 04:17 PM #5Junior Guru Wannabe
- Join Date
- Jan 2017
- Posts
- 34
Consider disaster recovery DR as a first step rather than high availability HA. If you customers truly want HA as in 99.999% availability, you can't get that on the cheap.
Suggestion: you work through the detail requirements with your customers, be it DR or HA, reach a Service Level Agreement with your customers, then plan and move forward accordingly.
-
01-31-2017, 04:26 PM #6Web Hosting Master
- Join Date
- Mar 2005
- Location
- Labrador, Canada
- Posts
- 988
Exactly.
'High-availabilty' might keep your sites online during a temporary outage. 'Disaster recovery' (backups) lets you recover from a hack (e.g., defaced site) or a user error (OMG I deleted something 2 days and just realized!)
High availabilty is akin to RAID. It's only temporary redundancy.
-
02-01-2017, 02:12 AM #7New Member
- Join Date
- Dec 2011
- Posts
- 3
Firstly, thank you all for your help here. I think we need both of these - backups and HA. We already do backups, thanks to DirectAdmin being able to generate them and we store them off-site.
Its more the HA side of things. Maybe I made the mistake of saying ‘cheap’. I heard that some hosting companies use VMWare to strip accounts across 2 or more servers, but its not something that I have come across. Any recommendations would be good.
Thanks again though. Really appreciate your time
Similar Threads
-
Is this right? What would you do as a host?
By betterthanyours in forum Web HostingReplies: 26Last Post: 11-20-2012, 01:50 PM -
What would you choose regarding server limitations?
By alexro in forum Web HostingReplies: 4Last Post: 02-24-2006, 04:27 AM -
What would you recommend as a Credit Card...
By Red Circle Web Hosting in forum Running a Web Hosting BusinessReplies: 7Last Post: 04-15-2001, 11:57 AM