What a denial of service attack is

A denial of service attack (DoS) is meant to make a service stop working. The attacker does not need to break in anywhere or steal anything: they load the website, the server or the internet line with more than it can handle, until real customers and employees can no longer connect. When the load comes from many sources at once, it is a distributed attack (Distributed Denial of Service, DDoS).

These sources are usually devices that were compromised in the past and are controlled remotely, computers, servers, cameras and home routers, linked together into one network (botnet). Their owners usually do not know. Because of the spread, you cannot block one address and be done: the traffic comes from everywhere, and some of it looks like normal browsing.

The motives vary. Some groups attack Israeli websites during periods of security or political tension (hacktivism). There is extortion, where the attacker threatens a shutdown unless money is paid (ransom DDoS). And there are cases where the outage is only a distraction from something else happening at the same time.

Where it hits: the line, the server or the application

To understand how to defend, it helps to tell the types of load apart. The difference between them also determines who can stop them.

  • Volume flooding (volumetric): an amount of traffic that fills the internet line itself. When the line is full, it does not matter how strong the server is, because legitimate traffic simply does not reach it.
  • Resource exhaustion at the protocol level (protocol attack): load that uses up the memory and connection table of network components, such as the firewall, the router or the load balancer, even when the line itself is not full.
  • Load on the application (Layer 7): requests that look like regular visitors but are aimed at heavy pages such as search, forms or a shopping cart. A website that was not built for this kind of load can crash even from a relatively small volume.

The target does not have to be the website. An attack on the office’s internet address can cut all employees off from cloud services, telephony and remote access (VPN), even when the website itself is hosted elsewhere and works as usual.

What raises the risk

  • A single internet line: when the whole office depends on one connection, flooding it stops all work.
  • A public service that sits on the office line: a web server, booking system or mail server hosted in the office exposes the line every employee works on.
  • Services open directly to the internet: management interfaces, remote access and servers with a public address. The wider the exposure to the internet (attack surface), the more targets there are.
  • No protection at the telecom or hosting provider: many providers offer traffic filtering, but it is not always on by default and not always included in the package.
  • An architecture without resilience: one server in one place, with no ability to spread the load or move elsewhere.
  • Public visibility: businesses associated with a controversial topic, public bodies and e-commerce sites during peak periods are more likely to be targeted.

There is also a financial risk that is talked about less. A cloud service configured to grow automatically with the load (auto scaling) can keep working during an attack and leave an unusual bill behind. It is worth setting cost caps and alerts in advance.

Early warning signs

A large attack is felt immediately, but sometimes there are signs before it. Attackers sometimes test the target’s response in short waves, and in extortion attacks a threat message comes first.

  • Sudden slowness of the website or of browsing in the office, without a technical change that explains it.
  • A sharp rise in traffic, especially from countries or networks that do not usually reach the website.
  • Unusual load on the firewall or the router, or alerts about an unusual number of connections.
  • Short, repeated outages at unexpected hours.
  • An email or message threatening a shutdown and demanding payment.

Not every load is an attack. A successful marketing campaign, a news story or a fault at the telecom provider can look similar. To tell the difference you need to know what normal traffic looks like (baseline), and that is only possible when there is ongoing monitoring and someone already knew the data before the day of the incident.

The impact on the organization

The immediate damage is to availability. A website that does not load takes no orders, a contact form receives no inquiries, and customers who cannot connect to the portal call the support line or move to a competitor. When the attack targets the office line, employees are cut off from email, cloud files, the accounting system and the phones that work over the internet (VoIP).

Availability is one of the pillars of information security, alongside confidentiality and integrity. That is why a denial of service attack is a full-fledged information security incident, even when no data is exposed. For a business that committed to customers to keep a service available, the outage can also become a contractual problem.

And there is what happens behind the scenes. When the whole team is busy getting the website back up, fewer people look at the other alerts. An attack like this is sometimes used as a smokescreen for an intrusion attempt, so after the load calms down you also need to check what else happened during those hours.

How to reduce the risk

Most protection against flooding cannot be built inside the office. It needs to sit somewhere with enough bandwidth to absorb the load and filter it before it reaches you.

  • Filtering at the provider (scrubbing): find out from the telecom provider and the hosting provider what protection exists, when it kicks in and who turns it on.
  • A content delivery network (CDN) and a web application firewall (WAF): for a public website, a layer like this absorbs most of the traffic, filters suspicious requests and hides the address of the original server (origin).
  • Request rate limiting: a cap on the number of requests from each source to the heavy pages, such as search, sign-in and forms.
  • Separating the website from the office: the website and public services are hosted outside the line employees work on, so an attack on one does not shut down the other.
  • A backup line (failover): a second internet connection, preferably from a different provider and over different infrastructure, with a switchover set up in advance.
  • Reducing exposure: close to the internet every service that does not have to be open, and move remote access to an identity-based solution without a classic VPN (ZTNA).
  • A scenario in the disaster recovery plan (DRP): contacts at the providers, an advance decision on who turns on the filtering, and an alternative channel for updating customers.

When it happens: working with the response team

The moment the website or the line goes down, the tendency is to try things: restart servers, change settings, block addresses by hand. In most cases this does not help, and sometimes it makes things worse. It is better to act in a clear order and bring in a cyber incident response team early:

  1. Diagnosis: confirm that it is an attack and not a fault at the provider or legitimate load, and identify which service was hit.
  2. Turning on filtering: contact the telecom or hosting provider and activate the agreed protection, or move the website to a filtering service.
  3. Keeping essential services running: switch to the backup line, work through another connection, and update customers on a channel that still works.
  4. Checking what happened in parallel: the response team goes over the system logs and the alerts from those hours, to make sure the outage did not hide an intrusion.
  5. Handling extortion: do not negotiate with the attacker and do not pay. Paying does not guarantee the attack will stop, and it marks the business as willing to pay.
  6. Documentation and reporting: record the timeline and the actions taken. The incident can be reported to the National Cyber Directorate, and if it turns out personal data was exposed, the duty to report to the Privacy Protection Authority is also reviewed.
  7. Lessons learned: after the incident is over, update the protections and the plan based on what was learned.

Where the IT company fits in

Protection against denial of service is spread across several parties: the telecom provider, the hosting provider, the CDN service, the office firewall and whoever manages the website. On the day of an attack, whoever holds the full picture knows who to call and what to ask for. An IT company that knows the network, the providers and the settings can act immediately, instead of wasting the first hours on inquiries.

Most of the work is done beforehand: mapping everything exposed to the internet, checking the protections already included in the provider contracts, separating public services from the office network, and a short drill of the scenario. This is part of an approach to information security in which availability is considered part of protection, not an operational topic handled separately.

Summary

  1. You know what is exposed to the internet, and which of it really has to be open.
  2. The website and public services are hosted separately from the office line, behind a CDN and a WAF.
  3. You have checked with the telecom and hosting providers what filtering exists and how to turn it on.
  4. There is a second internet line with a configured switchover.
  5. There is monitoring that shows what normal traffic looks like.
  6. The scenario is written into the recovery plan, including contacts and an alternative channel for customers.