IB Computer Science Topic A2.2.2: Function of Servers Notes

Servers: Machines Built to Answer One Kind of Request Well

1. A server is a computer system that provides a service, data or a resource to other computers, called clients, across a network.
2. Server is a role rather than a box: one physical machine can run a web server and a mail server at once, and one busy service can be spread across hundreds of machines.
3. The syllabus names six types, and each one is judged on the same four factors: function, scalability, reliability and security.

DNS Server: Turning Names into Addresses

1. Function
   1. Translates a human-readable domain name such as www.google.com into the numerical IP address a network actually routes to, such as 172.217.16.14.
   2. Devices communicate using IP addresses, so without DNS a user would have to memorise an address for every site they visit.
2. Scalability
   1. The name space is hierarchical, so a query for one domain is answered by the servers responsible for that branch rather than by one global machine.
   2. Adding a new domain adds load only to its own name servers, which is how DNS has scaled to billions of names.
3. Reliability
   1. Every domain is required to publish more than one name server, so one failure does not make the domain unreachable.
   2. Results are cached by resolvers and by the client itself, which cuts response time and keeps names resolving during a short outage.
4. Security
   1. Plain DNS has no way to prove an answer is genuine, so an attacker can reply with a false address and send users to a fake site.
   2. DNSSEC signs the records so a resolver can check the reply came from the real name server.

DHCP Server: Configuring Devices the Moment They Join

1. Function
   1. Assigns an IP address to a joining device automatically, along with the subnet mask, default gateway and DNS server it needs to use.
   2. Addresses are handed out as a lease with an expiry, so an address returns to the pool when a device leaves.
2. Scalability
   1. One pool of addresses can serve far more devices than it holds, because leases expire and are reused.
   2. A campus with thousands of transient laptops is configured by one server instead of by hand at every desk.
3. Reliability
   1. A failed DHCP server stops new devices joining, although devices holding a valid lease keep working until it expires.
   2. Larger networks run a second DHCP server with a split pool so that neither one is a single point of failure.
4. Security
   1. Reservations tie a fixed address to a known MAC address, so servers and printers always appear at the same place in firewall rules.
   2. A rogue DHCP server on the same network can hand out a false gateway address and route traffic through an attacker, so switches are configured to accept DHCP replies only from trusted ports.

File Server: One Shared Store Instead of Copies on Every Machine

1. Function
   1. Stores files centrally and serves them to many users, so everyone edits one copy rather than their own version.
   2. File locking stops two people writing to the same file at the same time and losing one set of changes.
2. Scalability
   1. Capacity grows by adding drives, and throughput grows by adding network bandwidth.
   2. The limit is usually the network link rather than the disks, because every read crosses the LAN.
3. Reliability
   1. RAID spreads data across several disks so the server keeps running when one disk fails.
   2. RAID protects against disk failure but not against deletion, so scheduled backups are still needed.
4. Security
   1. Access permissions decide which user or group can read, write or delete each folder.
   2. Encryption at rest keeps the data unreadable if a drive is stolen from the building.

Mail Server: Sending, Receiving and Storing Messages

1. Function
   1. Accepts outgoing mail over SMTP, stores messages for each account, and releases them to clients over IMAP or POP3.
   2. IMAP leaves the message on the server so several devices see the same mailbox, while POP3 downloads and usually removes it.
2. Scalability
   1. Load balancing spreads incoming connections across a cluster, so peak-hour traffic does not queue behind one machine.
   2. Storage grows separately from processing, because mailboxes keep growing even when the message rate does not.
3. Reliability
   1. A domain lists several mail servers in priority order, so a sender falls back to the next one when the first does not answer.
   2. Sending servers queue and retry for days rather than discarding mail, so a short outage delays messages instead of losing them.
4. Security
   1. Spam filters and antivirus scanning inspect messages before they reach a mailbox.
   2. SPF, DKIM and DMARC let a receiving server check that a message really came from the domain it claims, which is the main defence against forged sender addresses in phishing.

Proxy Server: An Intermediary That Filters, Caches and Hides

1. Function
   1. Sits between clients and the internet, forwarding requests on the client's behalf and returning the reply.
   2. Because every request passes through it, the proxy is the natural place to filter content, log usage and cache popular pages.
2. Scalability
   1. Caching means a page requested by 500 students is fetched from the internet once and served locally 499 times.
   2. That cuts the bandwidth the organisation has to buy as the number of users grows.
3. Reliability
   1. A single proxy is a bottleneck and a single point of failure, since nothing reaches the internet if it stops.
   2. Organisations run a pair of proxies with automatic failover so traffic switches to the second one without users noticing.
4. Security
   1. Outbound traffic leaves with the proxy's address, so an external site never sees the internal address of the client.
   2. Blacklists and whitelists enforce which sites are reachable, and the request log gives a record of who went where.

Web Server: Delivering Pages Over HTTP and HTTPS

1. Function
   1. Receives an HTTP or HTTPS request for a resource, builds or retrieves it, and returns it with a status code.
   2. Static files are read straight from disk, while dynamic pages are generated per request from a database.
2. Scalability
   1. A load balancer distributes requests across many identical web servers, so capacity grows by adding machines.
   2. This works because the servers hold no session state of their own, which is what lets any of them answer any request.
3. Reliability
   1. The load balancer health-checks each server and stops sending traffic to one that fails.
   2. Copies in more than one data centre keep the site up when a whole site loses power.
4. Security
   1. TLS encrypts the connection and proves the server's identity through a digital certificate, which is the difference between HTTP and HTTPS.
   2. Input validation on anything the page sends to a database is what stops SQL injection and cross-site scripting.

The Four Factors Behind Every Server Answer

1. Function
   1. Name the one job the server does, in the terms the syllabus uses, before saying anything else about it.
2. Scalability
   1. Say what grows when demand rises: machines behind a load balancer, disks in an array, or branches in a hierarchy.
3. Reliability
   1. Name the redundancy and say what a user actually experiences when one component fails.
4. Security
   1. Pair each threat with the specific countermeasure for that server, since a general answer about encryption and anti-malware earns little.