In a small structure, the file server is often the first server deployed. It has been dedicated to back up PCs or sharing documents. Its choice depends on security, availability, and performance constraints 6 point guide.
You are a VSE-SME, and your activity is growing. The computer system that you used until then without reaching saturation, is showing signs of weakness in the face of several recurring needs. It is therefore impossible to continue your industrial adventure without pooling your data, without connecting the PCs to each other more closely. Here is a roadmap - part one of our dossier - to deploy a file server, in 6 key steps. The answer in detail the three fundamental questions: how, how much, and why?
1 Objective: data sharing and backupsData sharing, possibly in a heterogeneous environment, is the primary motivation for a file server deployment. Each workstation under Windows, Mac OS, or Linux then sees the server as a disk drive. Several users thus work on the same documents, certainly not at the same time. Another advantage compared to local storage: the disk space is better exploited because it is shared. The server is also often used to centralize the backup of documents hosted by PC disks. Finally, a file server can also be used to share a printer.
2 Selection criteria: performance, security, scalabilityBetween a simple PC and a dedicated fault-tolerant server, the solutions are very disparate. The first selection criterion lies in the criticality of the data and the need to be able to access it without the slightest interruption. In fact, even on a very inexpensive product, a backup always makes it possible to protect the files. But in the event of a breakdown, the time to restore and restart can be considered prohibitive. It is then necessary to opt for an architecture tolerating the failure of one disk, therefore of Raid 1 type (each data is copied on two disks) or Raid 5 (the data is distributed over several disks).
Another criterion: the scalability directs the choice towards a solution allowing to add of disks. On the security side, it involves evaluating the need to finely manage access rights, possibly in an existing directory such as Active Directory. We will then distinguish, for each user or group, the authorized directories and files, as well as the read or write rights. As for performance, it depends on the power and memory of the server, as well as the speed of the disks and the network interface (100 Mbits / s or 1 Gbit / s).
3 Very economical solutions that are sometimes sufficientThere are practically free solutions, the easiest of which is to share a directory located on a user's PC. "Very often implemented in very small businesses, this architecture allows users to get used to the notion of the server", notes Franck Gotte, associate director of Deuzzi in the field. But the PC in question must remain on and maybe slowed down by the accesses of other users, while the rights management is very limited. “Another classic solution is to add a file sharing service to an existing application server,” he adds. You can also recycle an old PC by replacing its disk with a recent model and installing Linux and the free software Samba. This time, the access rights can be more finely allocated.
4 Dedicated file serversThe most elegant solution, but also the most expensive, is on the side of dedicated file servers - we speak of "NAS" ( Network Attached Storage) - of which there are many types. The most basic are boxes costing barely a hundred euros, in which we insert a standard hard drive. They are equipped with an Ethernet port which allows their connection to the network. But beware of performance in the event of concurrent access! In contrast, the most sophisticated file servers, whose cost starts at around a thousand euros, support dozens of simultaneous accesses and are scalable and tolerant to disk failure. Better, we change the defective disk without stopping the server, whose activity is however slowed down during the reconstitution of the lost data. In addition, these products automatically alert the administrator in the event of an impending failure or saturation. “Very often, in SMEs, no one monitors the file server, which causes data loss, ”argues Jorge Fernandez, responsible for strategic accounts at Iomega.
5 A deployment within the reach of non-IT specialistsFifteen minutes: this is the time limit for setting up file servers announced by the manufacturers. Concretely, it is generally sufficient to connect the server to the network (router or switch) then to assign an IP address to it, an operation that can be automated via the router's DHCP server or the automatic generation of a random address. The server is then visible to the PCs from which it is possible to perform configuration operations, such as rights management and the definition of alerts.
6 The essential safeguardIf the goal is to back up PCs, then educate users to periodically copy their critical files to the server or place them in local directories, which a backup utility will automatically copy to the server. Such tools are often provided by manufacturers. But if the goal is only to share files, then it is necessary to set up a backup procedure for the server itself, typically by copying all the data daily to another file server, to a tape cartridge drive or to a simple external USB hard drive.