Recommendations for SMB Server or Network Infrastructure
In the small and medium size business (SMB) segment it can be very difficult for businesses to justify the cost of having redundant servers and network switches to service multiple applications. The typical approach is to bundle all the applications within one or two physical servers with one network switch and firewall to serve both the servers /printers and users computers. With this setup and configuration, and in the event there is required recovery after a disaster situation, the downtime is huge and can be extremely costly to the business. Replacement parts could take days, even a week, not to mention that it could take days to re-install the operating system, applications and the restoration of the data using the data backup system on the server/ servers.
Recommendations/ best practices with regards to the setup for a small business (particularly where cost is a concern)
Implement one firewall, one network switch (or two of each) and two physical servers to balance out all network connections and applications. We also recommended that companies ensure they purchase the manufacturer’s support contract that provides same day response (7 days a week, 24 hours a day, with a 4 to 6 hour response time).
With regards to the server infrastructure we recommend that businesses virtualize the application servers. The best practice is to always have all applications spread out between multiple servers rather than having all applications bundled onto one server. The best way to accomplish this is to create one physical server configured as the hypervisor, (Microsoft, VMware or Citrix Zen server virtualization technology). For a small business it is more cost effective to use Microsoft due to operating licensing costs. With the one physical server being a hypervisor we can then build multiple virtual servers (cloud servers) running within the hypervisor. You might be thinking that with this strategy you are still running all the virtual servers on the same physical hardware. However, with having same day 7×24 onsite hardware support on the physical hypervisor while the virtual machines are still down, but they are just like files. The virtual server does not need to be re-built or re-installed. The hypervisor OS can be re-installed in a few hours.
The recovery strategy for the virtual machines becomes easier and much quicker. Along with nightly data backups for the data, and because the servers are virtual, we can create regular exports to an alternate storage device and if the virtual machine fails, we just delete it and import the files back to the hypervisor. Importing and exporting virtual machines is just like copying files from server to server.
A second recovery strategy that we put in place is having our systems administrators perform regular monthly maintenance such as Microsoft patches or application updates/upgrades. Our systems administrators create a snapshot of the virtual machine just before the maintenance takes place. The maintenance is applied and if there are any issues after the maintenance, the systems administrator has the ability to revert to the previous snap shot in minutes.
Nightly data backups
We recommend that there should always be a separate physical server performing nightly data backups, either to disk or to tape or both. The backup server would be used for storing the virtual machines exports and snapshots.