1

I have no experience in failover cluster area. I am building system for users that create new ftp site (currently Windows FTP). What I am thinking of is calculating server usage and choose server with lowest usage and deploy site to that server.

What I don't know is how Failover cluster works, this system will be used on failover cluster servers.

My example server list:

Server 1 - FTP, Web

Server 2 - FTP, Email

Server 3 - Web, Email

Server 4 - FTP, DNS

Server 5 - FTP, DNS

That takes me to point: Do I need to calculate (ask servers for usage) and deploy on lowest or should I pick first one and that server deploy widely ftp site?

What happens on adding ftp site to server 1? Is site automatically "replicated" to server 2, 4, 5? Thats my main question.

If server automatically say other servers in cluster to add/create site with those bindings or not.

Thanks, I hope you understand what I am asking :/

user1085907
  • 121
  • 1
  • Ok probably figured out solution: use shared configuration in windows IIS, mysql do automatic, dont know how for other solutions – user1085907 May 23 '17 at 18:08

1 Answers1

1

A failover cluster is a number of clustered nodes providing failover of service/VM/application to running partner node. It can be 2 or more clustered nodes. It depends on hypervisor/OS you go with and storage configuration is it converged (compute and storage host separated) or hyperconverged (when a node has both compute and storage resources).

It's highly recommended to familiarize yourself with Failover Clustering https://docs.microsoft.com/en-us/windows-server/failover-clustering/failover-clustering-overview

When it comes to plan an infrastructure, it would be advanced to define nines of availability before get the hardware. For next I would plan the number of VMs/services to run and calculate system requirements.

With regards to your example server list I would probably go hyperconverged take 2 servers with specifications to virtualize FTP, Web, Mail, DNS so each host could handle the entire production when partner host issued failure. With the implementation of shared nothing storage that mirror storage between hosts, I would have 4 to 5 nines uptime.

Mr. Raspberry
  • 3,918
  • 13
  • 32