hardware failover server configuration

Sandra1998

10/16/23, 11:06 AM

need to achieve redundancy for offline local small business Document managing system server and need to know which configuration is better considering zero down time and automated switching in failure situation without losing any data and hassle-free operation is there any configuration that can duplicate data in backup server simultaneously and take place of main server in failure case?

114

2 + 0

redundancy

Score:3

Server

batistuta09

11/3/23, 1:38 PM

Cluster with redundant shared storage is the approach, which allows to achieve near zero downtime and automated failover. The existing workload (Document Managing System, File Server,etc) should be virtualized and run inside cluster of your choice – Microsoft with Hyper-V, VMware vSphere, ovirt, Proxmox, etc. So, once one server will fail, virtual machines will automatically start on second server, that’s called high availability. Keep in mind, that at least 2 servers are required. As for the storage, I’d suggest looking at hyperconverged architecture here (where both servers have local storage, which is being replicated in real time between hosts), because redundant SAN with dual-controller will be an overkill for small business. Hypervisor’s vendors offer own solutions for local storage replication (Microsoft S2D, VMware vSAN), which are good, but may require some knowledge for deployment/managing and generate additional costs (separate VMware vSAN license, S2D requires Windows Datacenter). I have a good experience with Starwind vsan in small environments: full functional free version or commercial version, which comes with deployment assistance; easy of use without specific knowledge required; no strict hardware compatibility list.

Actually, cluster and high availability cannot replace backups, because each one is used for different purposes. Cluster – to make service available as soon as possible. Backups – to restore data, not only in case of server failure, but also after ransomware attack, human error, etc.

+ 2

Sandra1998

11/5/23, 10:29 AM

thanks for that, but what if main storage that virtual server installed on it faild? that what happened last time with bad sector problem on system drive. so i prefer 2 basic standalone servers like Dell T40. for storage i saw 2 synology NAS with High Availability feature is the simplest way to go.

BaronSamedi1958

1/13/24, 2:25 PM

No it's not. Host can fail, think about a controller failure within dual controller SAN (Nimble, EQL etc). All will just failover to the second alive host. With mirrored Syno you'll have crippled performance, need to buy extra units and deal with non-Enterprise Syno support which is legendary to answer on Taiwanese business hours.

Score:2

Server

Harry

10/16/23, 11:59 AM

I am assuming the server you're wishing to protect is a single standalone machine and remember this is offline/LAN only.

If there were a power outage, it would impact the whole site, in which case you should be looking into a UPS or generator for redundancy.

If you're looking at being able to take the server offline for maintenance and still have full (or partial) service available, then you would need to have a NAS for storage... you can mirror copies of data but this is problematic if you have an outage and changes are made to the secondary copy, you'd need to figure out which versions to use and recombine them.

... you could always use something like rsync to mirror to a hot-spare and have the storage there read-only but this feels like a lot of effort and maintenance for an unlikely situation when a NAS is a much easier setup to manage and pretty simple to mirror a NAS to another NAS for backup depending on which system you implement (take a look at TrueNAS).

Assuming you chose to move the data to a NAS, perhaps mounted over NFS on a VLAN onto your current server you would likely need a clone of this server always online and ready to go in case the primary went offline... this feels a little wasteful, it's perhaps better to share all load across two (or more) servers all the time and if one were to go down a load balancer would just remove it from the IP.

This suggestion would obviously require a load balancer, either hardware or software.

For a small office network, a hardware load balancer is about 30 quid from Amazon (NB this is 10/100, not GB) and a software option could be ipvs or if you have the hardware available take a look at pfSense which is free to download.

... and finally depending on your plans, your server could be virtualised, in which case you could run multiple instances of your office server load balanced on the same physical host.

I've made a boat load of assumptions about what you have in mind and your current setup, but i hope this helps.

+ 0

Elon Musk

I sit in a Tesla and translated this thread with Ai:

EN: hardware failover server configuration

Post an answer

Most people don’t grasp that asking a lot of questions unlocks learning and improves interpersonal bonding. In Alison’s studies, for example, though people could accurately recall how many questions had been asked in their conversations, they didn’t intuit the link between questions and liking. Across four studies, in which participants were engaged in conversations themselves or read transcripts of others’ conversations, people tended not to realize that question asking would influence—or had influenced—the level of amity between the conversationalists.