Mission Critical Services

In a Nutshell - CIW Course Section 3, Part B3, Chapter 1

Overview

In the context of Web servers there can be many dependencies. Services, without which the Web server cannot function. These services, which may be running on different servers can be considered to be the mission critical services.

Streaming Media servers, Database servers and Email servers may all be essential to the correct operation of the Web server. Mission critical servers and services need, as far as possible, to be available at all times. To this end, system security, system logging and fault tolerance, all have a part to play.

Network Security Services

Most company networks now require to be connected to the Internet. Without proper protection this can be a serious security risk. Firewalls, Gateways and Proxy Servers can help bridge this gap.

There are three main types of firewall:

1. Packet Filtering Firewall. Inspects packets and allows or denies access based on source and destination IP addresses and ports.
2. Application-Level Gateway. Operates at the application layer of the OSI/RM and is able to look deeper into packets.
3. Circuit-Level Gateway. Used, primarilly, for Network Address Translation.

If a breach of this security should occur then an Intrusion Detection System (IDS) can detect this and alert the administrator to suspicious activity.

System Logging

All enterprise servers will provide logging of, amongst other things, errors, user activity. and system alerts. The main location for logs in Windows 2000 is the "Event Viewer":

It can be seen from the tree-view in the left-pane, that the Event Viewer categorises messages into a number of groups. The event messages include informational messages, warnings and errors. Clearly the latter are the most significant. Double-clicking or right-clicking and selecting "Properties" on the event will display the Event Properties:

The event properties provide more information about the error or warning.

In Linux systems the logs are written to text files and may be stored in one of many locations. A typical log file is /var/log/messages which logs details of successful and failed login attempts.

Many services will require logging to be, explicitly, enabled before events will be written out to the log file.

System Performance and Fault Tolerance

The course notes talk about optimising server performance by implementing one or more of the following:

1. Upgrade existing hardware on your system.
2. Obtain additional systems to help distribute the load.
3. Make adjustments to the software.

To my mind, with the exception of number 3, this is not optimising server performance, it is improving server performance. Optimising is about maximising the performance with the hardware you have.

This might involve adjusting software settings to better utilise memory, or may need applications moved to alternative disks to improve access times or improve load balancing. Windows 2000 Performance Monitor lets you analyse memory, disk and network usage in an attempt to identify bottlenecks.

RAID disk systems can improve performance and/or provide redundancy for fault tolerance. By spreading data across several disks and/or multiple disk controllers access times can be improved and depending upon the RAID level chosen, disk failure can be tolerated as data can be reconstructed from the data on the remaining disks.

Server clustering is a method of load-balancing between servers. The clustering software will connect a client to the server in the group with the least load. If a server fails, or becomes unavailable for some other reason, the clustering software will detect this and process future connections accordingly. Examples of cluster servers are Citrix Metaframe, Microsoft Cluster Server, Sun Cluster and Apcon POWERSWITCH.

Parallel Processing clusters share the processing of individual tasks to improve the execution time of the task.