Prevent, backup and restore procedures for a web server

• Author: Cosmin Catalin Olteanu
• Pages: 483-489

Page483

1. Introduction

In the WEB integrated informatic systems era¹, reliability of servers is a big issue. If we have a real disaster plan for prevention and recovery of data loss and hardware failure, we can say that we have a trust worthy system and we can rely on it.

Every user should trust the system and should work with confidence that his work will be also available tomorrow.

2. Prevent, backup and restore a real disaster recovery plan

These days when everything is stored online on a server a real disaster recovery plan should be a must for every network administrator.

From my experience I would say that the first step is to prevent and than to make a detailed backup plan and a restore one too.

Let’s assume that we have a web generated application that is used as an integrated informatic system for a university campus.

We’ll go further and we’ll presume that everything is based on Linux (any distribution available, like Fedora 13, Gentoo, Debian etc.).

[SEE THE FIGURE IN THE ATTACHED PDF]

Fig. 1.Gentoo Logo.

Page484

[SEE THE FIGURE IN THE ATTACHED PDF]

Fig. 2.Fedora 13 Logo.

Data losses can happen any time due to many reasons and we have to manage that paper plan is not always equal with real plan:

♦ We can have accidental data loss

♦ We can have intentional data loss

♦ We have to deal with small budgets for components and not ideal ones etc.

I have to say that we have a few causes and I have discovered from my previous experience that this causes can be:

► electrical / power problems;

► failure of devices;

► bad coding

► Database bug’s etc.

a) electrical / power problems that can be avoided

I our country every issue about power supplier is a closed one because is just one supplier and we just do not have an alternative and usually when we encounter such problems the results are devastating.

Just a few years ago a mail server that I usually maintain just burn out because of a big overvoltage. The UPS and mainboard were fried.

After a few days (after I have replaced the components and the server was working again) I have tried to find an assurance company to have all the equipment assured but from 11 companies none of them could make an offer for electrical problems generated by supplier. The conclusion was just annoying: No one takes responsibility due to power failure. All must be done by lawyers, court law suites and time just could be extended for years until something is done.

All I could do in an environment where “time is money” was to separate the problem in two stages and deal with both of them:

1) problems generated through power cables (220v);

2) problems generated through small curents cables (UTP).

For the first problem the solution was to...